{"id":52,"date":"2020-09-03T16:38:52","date_gmt":"2020-09-03T13:38:52","guid":{"rendered":"http:\/\/acikkitap.com.tr\/oaek\/chapter\/bolum-8-dogal-dil-isleme-ve-ogrenme-analitigi\/"},"modified":"2020-09-03T16:38:52","modified_gmt":"2020-09-03T13:38:52","slug":"bolum-8-dogal-dil-isleme-ve-ogrenme-analitigi","status":"publish","type":"chapter","link":"https:\/\/acikkitap.com.tr\/oaek\/chapter\/bolum-8-dogal-dil-isleme-ve-ogrenme-analitigi\/","title":{"raw":"B\u00f6l\u00fcm 8 Do\u011fal Dil \u0130\u015fleme ve \u00d6\u011frenme Analiti\u011fi","rendered":"B\u00f6l\u00fcm 8 Do\u011fal Dil \u0130\u015fleme ve \u00d6\u011frenme Analiti\u011fi"},"content":{"raw":"\n

Danielle S. McNamara1<\/sup>, Laura K. Allen1<\/sup>, Scott A. Crossley2<\/sup>, Mihai Dascalu3<\/sup>, Cecile A. Perret4 <\/sup><\/span><\/span><\/p>\n1<\/sup>Psikoloji B\u00f6l\u00fcm\u00fc, Arizona Devlet \u00dcniversitesi, ABD <\/span><\/span>\n\n2<\/sup> Uygulamal\u0131 Dilbilim ve ESL B\u00f6l\u00fcm\u00fc, Georgia Devlet \u00dcniversitesi, ABD<\/span><\/span>\n\n3 <\/sup>Bilgisayar Bilimleri B\u00f6l\u00fcm\u00fc, B\u00fckre\u015f Politeknik \u00dcniversitesi, Romanya <\/span><\/span>\n\n4<\/sup> \u00d6\u011fretme ve \u00d6\u011frenme Bilim Enstit\u00fcs\u00fc, Arizona Devlet \u00dcniversitesi, ABD <\/span><\/span>\n\nDOI: 10.18608\/hla17.008<\/span><\/span>\n

\u00d6Z<\/h2>\nDil, bilgiyi anlamak ve ili\u015fki kurmak i\u00e7in bir kanal oldu\u011fundan e\u011fitim alan\u0131 i\u00e7in merkezi bir \u00f6neme sahiptir. Bu nedenle, \u00f6\u011frenme analiti\u011fi alan\u0131ndaki ara\u015ft\u0131rmac\u0131lar, dili do\u011fru ve etkili bir \u015fekilde analiz etmek i\u00e7in geli\u015ftirilen y\u00f6ntemlerden yararlanabilir. Do\u011fal dil i\u015fleme (DD\u0130) teknikleri b\u00f6yle bir yol sa\u011flayabilir. DD\u0130 teknikleri, belirli g\u00f6revlerle ilgili olarak dilin farkl\u0131 y\u00f6nlerinin bilgi i\u015flemsel analizlerini sa\u011flamak i\u00e7in kullan\u0131l\u0131r. Bu b\u00f6l\u00fcmde, yazarlar s\u00f6ylemin anla\u015f\u0131lmas\u0131 i\u00e7in kullan\u0131labilecek bir\u00e7ok DD\u0130 ara\u00e7lar\u0131 ve bu e\u011fitim ara\u00e7lar\u0131n\u0131n baz\u0131 uygulamalar\u0131n\u0131 tart\u0131\u015fmaktad\u0131r. Bu ara\u00e7lar\u0131n ana oda\u011f\u0131, insanlar ve bilgisayarlar aras\u0131ndaki etkile\u015fimi veya insan-bilgisayar etkile\u015fimini sa\u011flamak i\u00e7in insan dili girdisinin otomatik olarak yorumlanmas\u0131d\u0131r. Bu nedenle, ara\u00e7lar metinleri anlamak i\u00e7in \u00f6nemli olan tutarl\u0131l\u0131k, s\u00f6zdizimsel karma\u015f\u0131kl\u0131k, kelime \u00e7e\u015fitlili\u011fi ve anlamsal benzerlik gibi \u00e7e\u015fitli dil \u00f6zelliklerini \u00f6l\u00e7er. Yazarlar, b\u00f6l\u00fcm\u00fc DD\u0130 ara\u00e7lar\u0131n\u0131 (yani, Ak\u0131ll\u0131 \u00d6\u011fretim Sistemleri, KA\u00c7D'ler ve Bilgisayar Destekli \u0130\u015fbirlikli \u00d6\u011frenme) kullanan bilgisayar tabanl\u0131 \u00f6\u011frenme ortamlar\u0131 ve bu t\u00fcr ara\u00e7lar\u0131n gelecekteki ara\u015ft\u0131rmalarda nas\u0131l kullan\u0131labilece\u011fi tart\u0131\u015fmas\u0131yla sonland\u0131rmaktad\u0131r.<\/span>\n\nAnahtar Kelimeler<\/b>:Do\u011fal dil i\u015fleme (DD\u0130), dil, bilgi i\u015flemsel dil bilimi, dil \u00f6zellikleri, otomatik yazma de\u011ferlendirmesi, ak\u0131ll\u0131 \u00f6\u011fretici sistemleri, BD\u0130\u00d6, KA\u00c7D<\/span>\n

Dil d\u00fc\u015f\u00fcncelerimizi d\u0131\u015far\u0131ya yans\u0131tman\u0131n bir arac\u0131d\u0131r. Kendimizi ba\u015fkalar\u0131na ifade etmemize, d\u00fcnyay\u0131 manip\u00fcle etmemizi ve \u00e7evrede bulunan nesneleri etiketlememizi sa\u011flar. Dil, d\u00fc\u015f\u00fcncelerimizi i\u00e7sel olarak in\u015fa etmemizi ve yeniden yap\u0131land\u0131rmam\u0131z\u0131 sa\u011flar; d\u00fc\u015f\u00fcncelerimizi temsil edebilir ve onlar\u0131 d\u00f6n\u00fc\u015ft\u00fcrmemize imkan tan\u0131r. Sosyal deneyimler kurmam\u0131z\u0131 ve \u015fekillendirmemizi sa\u011flar. Dil, d\u00fcnyay\u0131 anlamak ve etkile\u015fimde bulunmak i\u00e7in bir kanal sa\u011flar.<\/p>\n

Dil ya\u015fam\u0131m\u0131zda, d\u00fc\u015f\u00fcncelerimizde, ileti\u015fimimizde, okuduklar\u0131m\u0131zda ve yazd\u0131klar\u0131m\u0131zda ve di\u011ferleriyle etkile\u015fimlerimizde her yerdedir. Dil, e\u011fitim i\u00e7in ayn\u0131 \u015fekilde merkezidir. \u00d6\u011fretenler olarak hedefimiz, \u00f6\u011frencilere yeni bilgileri \u00f6\u011frenme, bunlar\u0131 \u00f6z\u00fcmseme ve b\u00fct\u00fcnle\u015ftirme f\u0131rsat\u0131na sahip olacak \u015fekilde bilgileri iletmektir. \u00d6\u011frenciler bilgiyi iletmek i\u00e7in kullan\u0131lan dili anlama ve daha sonra bu bilgileri bireyler olarak, gruplar halinde, birbirleriyle ve \u00f6\u011fretenlerle e\u015fg\u00fcd\u00fcm halinde in\u015fa etmeyi bildikleri \u015feylerle ili\u015fkilendirmekle g\u00f6revlidirler.<\/p>\n

Dil, ya\u015fam\u0131m\u0131zda ve e\u011fitimde \u00f6nemli roller oynar ve bu nedenle bu rolleri ve \u00e7\u0131kt\u0131lar\u0131 tan\u0131mak ve anlamak \u00f6nemlidir. Metin ve s\u00f6ylem analizi, dil kullan\u0131m\u0131yla ilgili karma\u015f\u0131k s\u00fcre\u00e7leri anlamak i\u00e7in bir yol sa\u011flar. S\u00f6ylem analistleri, yaz\u0131l\u0131 metin ve s\u00f6zl\u00fc s\u00f6ylem i\u00e7erisindeki yap\u0131lar\u0131, \u00f6r\u00fcnt\u00fcleri ve onlar\u0131n davran\u0131\u015flarla, psikolojik s\u00fcre\u00e7lerle, bili\u015fsel ve sosyal etkile\u015fimlerle ili\u015fkilerini sistematik olarak inceler. Nitekim, metin ve s\u00f6ylem analizi dil hakk\u0131nda bir bilgi hazinesi sa\u011flam\u0131\u015ft\u0131r.<\/p>\n

Geleneksel olarak, s\u00f6ylem analizi zahmetlidir. \u0130lk olarak, \u00f6rne\u011fin, anlaml\u0131 dil birimleri tan\u0131mlan\u0131r ve b\u00f6l\u00fcmlere ayr\u0131l\u0131r (\u00f6r. t\u00fcmceler, ifadeler) ve sonra uzmanlar bu birimleri kodlar (yani, belirli bir analiz i\u00e7in). Ard\u0131ndan, bu dil birimlerinin do\u011fas\u0131 ve \u00e7\u0131kt\u0131lar\u0131 aras\u0131ndaki potansiyel ili\u015fkiler de\u011ferlendirilir. Bireyler aras\u0131nda binlerce ifade ve al\u0131\u015fveri\u015fin oldu\u011fu b\u00fcy\u00fck veri d\u00fcnyas\u0131nda, dili elle-kodlama neredeyse imk\u00e2ns\u0131zd\u0131r. B\u00fcy\u00fck veri derlemleri, dili daha geni\u015f ve daha anlaml\u0131 bir \u00f6l\u00e7ekte anlamak i\u00e7in kap\u0131lar\u0131 a\u00e7ar ancak s\u00f6ylem analizine geleneksel yakla\u015f\u0131mlar uygulanabilir ve elveri\u015fli de\u011fildir. Do\u011fal dil i\u015flemeden<\/i><\/span> (DD\u0130) bir \u00e7\u00f6z\u00fcm elde edilebilir.<\/p>\n

DD\u0130, insan dilinin bilgisayar dili kullanarak analizidir ve s\u00f6ylem \u00e7\u00f6z\u00fcmlemesini otomatikle\u015ftirmenin yolunu g\u00f6sterir. DD\u0130 terimi, bilgisayar dillerinin kullan\u0131m\u0131 ve analizinin aksine, do\u011fal insan dilinin analizi oldu\u011fu i\u00e7in kullan\u0131lm\u0131\u015ft\u0131r. Do\u011fal dili i\u015flemek i\u00e7in \u00e7e\u015fitli otomatik ara\u00e7lar kullan\u0131labilir. Nitekim, DD\u0130 ara\u00e7lar\u0131n\u0131n say\u0131s\u0131 ve g\u00fcc\u00fc 1990'lar\u0131n ortas\u0131ndan bu yana d\u00fczenli olarak artm\u0131\u015ft\u0131r (Jurafsky ve Martin, 2000, 2008). Dolay\u0131s\u0131yla \u00f6\u011frenme analiti\u011fi ve veri madencili\u011fi alan\u0131ndaki etkileri ve kullan\u0131mlar\u0131, katlamal\u0131 olarak olmasa da giderek artmaktad\u0131r. Bu b\u00f6l\u00fcm, ara\u015ft\u0131rmac\u0131lar ve e\u011fitimciler i\u00e7in dili bilgi i\u015flemsel olarak analiz etmek i\u00e7in mevcut olan ve \u00f6zellikle de e\u011fitim alan\u0131ndaki kullan\u0131mlar\u0131na odaklanan \u00e7e\u015fitli ara\u00e7lar\u0131 a\u00e7\u0131klamaktad\u0131r.<\/p>\n\n

DO\u011eAL D\u0130L \u0130\u015eLEME<\/h2>\n

Bilgii\u015flemsel dilbilim modelleri geli\u015fimine odaklanan bir disiplindir. DD\u0130 ara\u00e7 ve teknikleri genellikle bilgi i\u015flemsel dilbilimi alan\u0131nda geli\u015ftirilen teoriler, modeller ve algoritmalar taraf\u0131ndan y\u00f6nlendirilir ancak DD\u0130 ara\u00e7lar\u0131n\u0131n temel amac\u0131 insan dili girdilerinin otomatik olarak yorumlanmas\u0131d\u0131r. B\u00f6yle bir \u00e7aba, dilbilim, bilgisayar bilimi, psikoloji ve e\u011fitim gibi disiplinleri birle\u015ftiren disiplinleraras\u0131 bak\u0131\u015f a\u00e7\u0131lar\u0131 \u00fczerinde durur. DD\u0130'nin Turing'e (1950) dayanan bir ge\u00e7mi\u015fi olsa da mevcut DD\u0130 algoritmalar\u0131n\u0131n \u00e7o\u011fu, DD\u0130 ara\u00e7lar\u0131 ve veri madencili\u011finin birle\u015fimi kullan\u0131larak geli\u015ftirilmi\u015ftir. Bilgisayar veya veri bilimcileri taraf\u0131ndan s\u0131kl\u0131kla kullan\u0131lan DD\u0130 yaz\u0131l\u0131m\u0131 ile bu b\u00f6l\u00fcmde sunulan ara\u00e7lar aras\u0131nda en ba\u015ftan net bir ayr\u0131m yap\u0131lmal\u0131d\u0131r. DD\u0130'ye ili\u015fkin ara\u015ft\u0131rmalar\u0131n b\u00fcy\u00fck bir \u00e7o\u011funlu\u011fu y\u00fczeysel seviyede metin i\u015flemeye odaklanm\u0131\u015ft\u0131r (\u00f6r. makine \u00e7evirisi) ve mevcut ara\u00e7lar sonu\u00e7 olarak do\u011fru kelime -ve c\u00fcmle- d\u00fczeyi metin i\u015flemenin temel rol\u00fcn\u00fc vurgulamaktad\u0131r. Bu b\u00f6l\u00fcmdeki amac\u0131m\u0131z, \u00f6\u011frenme analiti\u011fi ba\u011flam\u0131nda DD\u0130'ye odaklanmakt\u0131r. Bu nedenle, bu y\u00fczey d\u00fczeyinde g\u00f6revlerin \u00f6tesine ge\u00e7en, kelime dizinlerini hesaplamak i\u00e7in geli\u015ftirilen ve e\u011fitim ba\u011flam\u0131nda daha \u00f6nemli olabilecek bilgileri sunan ara\u00e7lara odaklan\u0131yoruz. Bilhassa, birden fazla metin d\u00fczeyi hakk\u0131nda bilgi sa\u011flayan bir DD\u0130 teknikleri alt k\u00fcmesini betimliyoruz. Bu ara\u00e7lar s\u00f6ylemdeki s\u00f6zc\u00fcklerden ba\u015flar, belirli kelime \u00f6zelliklerini \u00e7\u0131kar\u0131r ve ard\u0131ndan anlam yap\u0131s\u0131n\u0131 ve s\u00f6ylem yap\u0131s\u0131n\u0131 dikkate alarak veri s\u00f6zl\u00fc\u011f\u00fcn\u00fcn \u00f6tesine ge\u00e7er. Amac\u0131m\u0131z, mevcut t\u00fcm y\u00f6ntemlere genel bir bak\u0131\u015ftan ziyade, birka\u00e7 ortak tekni\u011fe dair \u00f6rnekler sunmakt\u0131r. Bu y\u00f6ntemleri,do\u011frudan analiz birimleri olarak kelimelere odaklananlar ve kelimelerin \u00f6zelliklerine odaklananlar olarak grupland\u0131r\u0131yoruz.<\/p>\n\n

Kelimeler<\/h3>\n

DD\u0130'ye y\u00f6nelik bir yakla\u015f\u0131m, dilde kullan\u0131lan kelimeleri do\u011frudan analiz etmektir. \u00d6rne\u011fin, bir metindeki belirli s\u00f6zc\u00fck t\u00fcrlerinin g\u00f6r\u00fclme s\u0131kl\u0131\u011f\u0131n\u0131 hesaplamak, \u00e7e\u015fitli ba\u011flamlarda kullan\u0131lan dilin do\u011fas\u0131 ve amac\u0131na dair iyi bir y\u00f6ntem olu\u015fturabilir. Bu genellikle \"s\u00f6zc\u00fck \u00e7antas\u0131\" yakla\u015f\u0131m\u0131 olarak adland\u0131r\u0131l\u0131r. Bu yakla\u015f\u0131m\u0131 kullanan ara\u00e7lardan biri, Pennebaker ve meslekta\u015flar\u0131, taraf\u0131ndan geli\u015ftirilen Dilbilimsel Sorgu ve Kelime Say\u0131s\u0131 (DSKS) sistemidir (Pennebaker, Booth ve Francis, 2007; Pennebaker, Boyd, Jordan ve Blackburn, 2015; bk. http: \/\/liwc.wpengine.com). DSKS'nin 2007 s\u00fcr\u00fcm\u00fc kabaca 80 kelime kategorisi sa\u011flar, fakat ayn\u0131 zamanda bu kelime kategorilerini daha geni\u015f boyutlarda grupland\u0131r\u0131r. Daha geni\u015f boyutlar\u0131n \u00f6rnekleri dil bi\u00e7imleri (\u00f6r. zamirler, ge\u00e7mi\u015f zamandaki kelimeler, ters ifadeler), sosyal s\u00fcre\u00e7ler, duygusal s\u00fcre\u00e7ler ve bili\u015fsel s\u00fcre\u00e7lerdir. \u00d6rne\u011fin, bili\u015fsel s\u00fcre\u00e7ler, i\u00e7g\u00f6r\u00fc (\u00f6r. d\u00fc\u015f\u00fcn, bil, g\u00f6z \u00f6n\u00fcne al), nedensellik (\u00f6r. \u00e7\u00fcnk\u00fc, etki, dolay\u0131s\u0131yla) ve kesinlik (\u00f6r. her zaman, asla) gibi alt kategorileri i\u00e7erir. DSKS, her bir kelime kategorisine ait kelimelerin say\u0131s\u0131n\u0131 sayar ve kategorideki kelimelerin say\u0131s\u0131n\u0131 metindeki kelimelerin toplam say\u0131s\u0131na b\u00f6lerek bir oran sunar.<\/p>\n

Benzer bir yakla\u015f\u0131m, karakterlerin veya kelimelerin gruplar\u0131 gibi n-gramlar\u0131n\u0131 tan\u0131mlamakt\u0131r; burada n, gruba d\u00e2hil edilen gramlar\u0131n say\u0131s\u0131n\u0131 belirtir (\u00f6r. \u0130ki gramlar, iki kelimeli gruplara at\u0131fta bulunur). N-gram analizleri, metinlerdeki kelime dizilimlerin olas\u0131l\u0131k da\u011f\u0131l\u0131mlar\u0131n\u0131 hesaplar ve bir metin grubuna ortak olan veya belirli bir metin veya metin dizileri i\u00e7in farkl\u0131 olan kelimeler hakk\u0131nda bilgi sa\u011flayabilir (\u00f6r. Jarvis vd., 2012). N-gram analizlerinin avantajlar\u0131, basitlikleri ve bir metnin spesifik i\u00e7eri\u011fi, bir metnin dil ve s\u00f6zdizimsel \u00f6zellikleri ve bu \u00f6zellikler aras\u0131ndaki ili\u015fkiler hakk\u0131nda bilgi sa\u011flama potansiyelini i\u00e7erir (Crossley ve Louwerse, 2007).<\/p>\n\n

Kelimelerin \u00d6zellikleri<\/h3>\n

Kelimelerin olu\u015fumunu ve kelime gruplar\u0131n\u0131 hesaplamak metnin a\u00e7\u0131k i\u00e7eri\u011fini g\u00f6z \u00f6n\u00fcne al\u0131r. Alternatif bir yakla\u015f\u0131m, bir metindeki kelimelerin ve c\u00fcmlelerin \u00f6zelliklerinin hesaplanmas\u0131n\u0131 i\u00e7erir. Bu t\u00fcr bir teknik, kelimelerin arkas\u0131ndaki gizli anlam\u0131 elde etmek i\u00e7indir (McNamara, 2011). Bunu yapmak i\u00e7in \u00e7ok say\u0131da algoritma olsa da en iyi bilinen ve belki de ilk olan\u0131 \u00f6rt\u00fck semantik analiz (\u00d6SA; Landauer ve Dumais, 1997; Landauer, McNamara, Dennis ve Kintsch, 2007; bk. lsa.colorado.edu) dir. \u00d6SA 1990'lar\u0131n ortalar\u0131nda ortaya \u00e7\u0131kt\u0131, b\u00fcy\u00fck metin g\u00f6vdelerinden semantik anlam \u00e7\u0131karmak ve b\u00fcy\u00fck ve k\u00fc\u00e7\u00fck metin \u00f6rneklerini semantik benzerliklerle kar\u015f\u0131la\u015ft\u0131rmak i\u00e7in bir ara\u00e7 sundu. Bu yakla\u015f\u0131m DD\u0130'de bir devrim yaratmak i\u00e7in benzersiz bir potansiyel sa\u011flad\u0131. \u00d6SA, geni\u015f bir belge k\u00fcmesinde kelimelerin var olu\u015funu temsil eden bir matrisi s\u0131k\u0131\u015ft\u0131rmak (yani \u00e7arpanlara ay\u0131rmak) i\u00e7in tekil de\u011fer ayr\u0131\u015ft\u0131rmas\u0131 kullanan matematiksel, istatistiksel bir tekniktir. \u00d6SA'y\u0131 y\u00f6nlendiren temel varsay\u0131m kelimelerin anlamlar\u0131n\u0131n onlarla birlikte olanlar taraf\u0131ndan yakaland\u0131\u011f\u0131 idi. \u00d6rne\u011fin, \"veri\" kelimesi, \"hesaplamalar\", \"madencilik\", \"bilgisayar\" ve \"matematik\" gibi ayn\u0131 i\u015flevsel ba\u011flamdaki kelimelerle b\u00fcy\u00fck \u00f6l\u00e7\u00fcde ili\u015fkilendirilecektir. Bu kelimeler veri ile ayn\u0131 anlama gelmemektedir. Daha do\u011frusu, bu kelimeler genellikle benzer ba\u011flamlarda olu\u015ftu\u011fu i\u00e7in verilerle ilgilidir. \u00d6SA kelimeler, c\u00fcmleler ve paragraflar aras\u0131ndaki semantik benzerliklerin hesaplanmas\u0131n\u0131 sa\u011flayarak, metinde anlam simulasyonunun kap\u0131lar\u0131n\u0131 a\u00e7t\u0131 (McNamara, 2011). \u00d6SA, basit kelime \u00f6rt\u00fc\u015fme \u00f6nlemlerinin yeterli olmad\u0131\u011f\u0131 bir problem olan anlaml\u0131l\u0131k sorununu (\u00f6r. bir metnin ba\u015fka bir metinle veya \u00e7ekirdek bir kavramla ne derece alakal\u0131 oldu\u011fu) ba\u015far\u0131yla ele alan ilk kelime temelli yakla\u015f\u0131m olarak d\u00fc\u015f\u00fcn\u00fclebilir. \u00d6SA'n\u0131n \u00f6tesine ge\u00e7en \u00e7ok say\u0131da yakla\u015f\u0131m olsa da (genel bir bak\u0131\u015f i\u00e7in bk. McNamara, 2011), \u00d6SA, kelime anlam\u0131n\u0131 modellemek ve anlam bilim ve metin uyumu bak\u0131m\u0131ndan i\u00e7g\u00f6r\u00fc sa\u011flamak i\u00e7in bir\u00e7ok ba\u011flamda kullan\u0131lan ortak bir yakla\u015f\u0131m olmaya devam etmektedir (\u00f6r. Landauer vd., 2007; McNamara, Graesser, McCarthy ve Cai, 2014).<\/p>\n

Dilin bariz bir \u00f6zelli\u011fi anlam\u0131d\u0131r ancak konu\u015fma b\u00f6l\u00fcmleri (\u00f6r. fiil, isim), s\u00f6zdizimi, psikolojik y\u00f6nler (\u00f6r. somutluk, anlaml\u0131l\u0131k) ve aras\u0131ndaki ili\u015fkiler, metindeki fikirler (\u00f6r. uyum) gibi bir \u00e7ok di\u011fer \u00f6zellikler dilbilimsel analizlerden t\u00fcretilebilir. Coh-Metrix, ilk olarak 2003 y\u0131l\u0131nda piyasaya s\u00fcr\u00fclen, metnin dil, psikolojik ve semantik \u00f6zelliklerini \u00e7\u0131karmak i\u00e7in dil hakk\u0131nda bir\u00e7ok bilgi kayna\u011f\u0131 kullanan otomatik bir dil analiz arac\u0131 \u00f6rne\u011fidir (McNamara vd., 2014; cohmetrix.com). Coh-Metrix, \u0130ngilizce dili ile ilgili bilgileri \u00d6SA, MRC Psikodilbilimsel Veri Taban\u0131, WordNet ve CELEX gibi kelime s\u0131kl\u0131\u011f\u0131 dizinleri ile s\u00f6zdizimsel ayr\u0131\u015ft\u0131r\u0131c\u0131lar gibi \u00e7e\u015fitli kaynaklardan uyarlar ve birle\u015ftirir. \u00d6rne\u011fin, MRC Psikodilbilimsel Veri Taban\u0131, kelimeler hakk\u0131nda psikodilbilimsel bilgi sa\u011flar (Wilson, 1988) ve WordNet, kelimelerin dil ve anlam \u00f6zellikleri ve ayr\u0131ca kelimeler (Miller, Beckwith, Fellbaum, Gross ve Miller, 1990) aras\u0131ndaki anlamsal ili\u015fkileri de sa\u011flar. Coh-Metrix ayr\u0131ca, yaz\u0131l\u0131 veya s\u00f6zl\u00fc metinlerin \u00e7ok boyutlu bir analiz \u00fcretmek i\u00e7in s\u00f6zc\u00fck s\u0131kl\u0131\u011f\u0131 ve c\u00fcmle uzunlu\u011fu gibi metin kalitesinin basit \u00f6zellikleri, tutarl\u0131l\u0131k ve s\u00f6zdizimsel karma\u015f\u0131kl\u0131k gibi daha karma\u015f\u0131k \u00f6zellikler sayesinde dilin \u00e7e\u015fitli y\u00f6nleriyle ilgili dil indekslerini de hesaplar. (McNamara, Ozuru, Graesser ve Louwerse, 2006). Coh \u2013 Metrix, tan\u0131mlay\u0131c\u0131 dizinlerle (\u00f6r. kelimelerin uzunlu\u011fu, c\u00fcmleler, paragraflar) metnin g\u00f6rece basit bir nitelemesini sa\u011flayabilir. Ek olarak, bir metnin kalitesini ve okunabilirli\u011fini tan\u0131mlayan \u00e7e\u015fitli karma\u015f\u0131k indisler sunar. Bu dizinler aras\u0131nda, anlat\u0131, referans uyumu, s\u00f6zdizimsel basitlik, s\u00f6zc\u00fck somutlu\u011fu ve derin bir uyum d\u00e2hil olmak \u00fczere be\u015f Coh-Metrix Metin Kolayl\u0131k Bile\u015feni bulunmaktad\u0131r (Graesser, McNamara ve Kulikowich, 2011; Jackson, Allen ve McNamara, 2016; bk. metrix.commoncoretera.com).<\/p>\n

Coh-Metrix, otomatik dil analizini herkese a\u00e7\u0131k hale getirerek dil ve s\u00f6ylem anlay\u0131\u015f\u0131m\u0131z\u0131 b\u00fcy\u00fck \u00f6l\u00e7\u00fcde etkilemi\u015ftir. Coh-Metrix birden fazla dil \u00f6l\u00e7\u00fcs\u00fc sa\u011flarken, Coh-Metrix'in \u00f6ncelikli ve \u00f6zg\u00fcn oda\u011f\u0131 metinde uyum \u00f6l\u00e7\u00fctleri sa\u011flamak olmu\u015ftur. Uyum, c\u00fcmle (yani, yerel uyum), paragraflar (yani, k\u00fcresel uyum) ve metnin geneli(\u00f6r. s\u00f6zc\u00fck \u00e7e\u015fitlili\u011fi) gibi daha b\u00fcy\u00fck b\u00f6l\u00fcmlerin, kelimeler ve anlam c\u00fcmleleriyle \u00f6rt\u00fc\u015fmesidir. Son derece yararl\u0131 olsa da Coh-Metrix'in kolayl\u0131k ve geni\u015f uyum indisleri \u00f6l\u00e7\u00fcm\u00fc ile ilgili baz\u0131 eksiklikleri vard\u0131r. \u0130lk olarak, metnin toplu i\u015flenmesine izin vermez ve kullan\u0131c\u0131n\u0131n sabit diskinde durmaz (ve bu nedenle internet ba\u011flant\u0131s\u0131na ve harici bir sunucuya ba\u011fl\u0131d\u0131r). \u0130kincisi, Coh-Metrix uyum indislerileri genel olarak k\u00fcresel uyuma (\u00f6r. bir metnin \u00e7e\u015fitli b\u00f6l\u00fcmleri aras\u0131nda semantik \u00f6rt\u00fc\u015fme) de\u011fil, yerel ve genel metin uyumuna (yani ortalama c\u00fcmle \u00e7ak\u0131\u015fmas\u0131, s\u00f6zc\u00fck \u00e7e\u015fitlili\u011fi) odaklan\u0131r. Bu nedenle, Metin Uyumunun Otomatik Analizi Arac\u0131 (MUOAA) ve Basit Do\u011fal Dil \u0130\u015fleme Arac\u0131 (BDD\u0130A) bu bo\u015fluklar\u0131 ele almak i\u00e7in geli\u015ftirilmi\u015ftir (Crossley, Allen, Kyle ve McNamara, 2014; Crossley, Kyle ve McNamara, bas\u0131m a\u015famas\u0131nda; http:\/\/www.kristopherkyle.com\/taaco.html). MUOAA yerel olarak (bilgisayara) kurulur (bir internet aray\u00fcz\u00fcne k\u0131yasla), metin dosyalar\u0131n\u0131n toplu olarak i\u015flenmesini sa\u011flar ve yerel, k\u00fcresel ve genel metin uyumu ile ilgili 150'den fazla indis i\u00e7erir. Benzer \u015fekilde, BDD\u0130A yerel olarak kurulur ve toplu metin i\u015flemeye izin verir. Bununla birlikte, BDD\u0130A, MUOAA'dan farkl\u0131d\u0131r; metinlerin bir\u00e7ok y\u00f6n\u00fcyle ilgili bilgileri hesaplamak i\u00e7in \"s\u00f6zc\u00fck \u00e7antas\u0131\" yakla\u015f\u0131m\u0131n\u0131 kullan\u0131r. Ek olarak, ara\u00e7 esnektir ve ara\u015ft\u0131rmac\u0131lara ek analizler sunmak i\u00e7in kendi s\u00f6zc\u00fck kategorilerini eklemelerini sa\u011flar.<\/p>\n

Serbest\u00e7e eri\u015filebilen bir DD\u0130 arac\u0131n\u0131n bir ba\u015fka \u00f6rne\u011fi, S\u00f6zc\u00fcksel Kapsaml\u0131l\u0131\u011f\u0131n Otomatik Analizi Arac\u0131d\u0131r (SKOAA; Kyle ve Crossley, 2015; http:\/\/www.kristopherkyle.com\/taales.html). SKOAA, bir metinde mevcut s\u00f6zc\u00fcksel karma\u015f\u0131kl\u0131\u011f\u0131n seviyesi hakk\u0131nda kapsaml\u0131 bilgi sa\u011flamaya odaklan\u0131r. Bu t\u00fcr bir analiz \u00f6nemlidir, \u00e7\u00fcnk\u00fc bir metnin s\u00f6zc\u00fcksel talepleri hakk\u0131nda bilgi sa\u011flamas\u0131n\u0131n yan\u0131 s\u0131ra, metnin yazar\u0131n\u0131n s\u00f6zc\u00fcksel bilgisi ile ilgili potansiyel bilgi de sa\u011flar (Kyle ve Crossley, 2015). SKOAA, bir metinde kullan\u0131lan s\u00f6zc\u00fck bilgisinin geni\u015fli\u011fini ve derinli\u011fini de\u011ferlendirmek i\u00e7in 130'un \u00fczerinde klasik ve yeni geli\u015ftirilen s\u00f6zc\u00fck indisini hesaplar. Bu ara\u00e7 h\u0131zl\u0131, g\u00fcvenilir ve \u00fccretsiz indirilebilir. SKOAA i\u00e7in al\u0131nacak \u00f6nlemler aras\u0131nda kelime s\u0131kl\u0131\u011f\u0131, kelime ve kelime ailesi \u00e7e\u015fitleri, n-gram, akademik listeler ve psikodilbilimsel unsurlar\u0131 dikkate alan kelime bilgisi indisleri bulunmaktad\u0131r (Kyle ve Crossley, 2015). Bu indisler toplu olarak, kelime se\u00e7imlerinin metindeki karma\u015f\u0131kl\u0131\u011f\u0131 hakk\u0131nda geni\u015f bilgi sa\u011flar.<\/p>\n

Dascalu, McNamara, Crossley ve Trausan-Matu (2016), ayr\u0131ca, bireysel kelimelerin \u00f6\u011frenme oran\u0131n\u0131n, \u00f6\u011frenenin dille ilgili deneyiminin bir fonksiyonu olarak hesapland\u0131\u011f\u0131 kelime karma\u015f\u0131kl\u0131\u011f\u0131n\u0131 tahmin etmek i\u00e7in hesaplanan bir model olan Maruz Kalma Ya\u015f\u0131'n\u0131 da (MKY) tan\u0131tm\u0131\u015ft\u0131r. Pearson'un kelime olgunlu\u011fu hesaplamas\u0131n\u0131n tersine (Landauer, Kireyev ve Panaccione, 2011), MKY, zaman i\u00e7inde veya daha \u00f6zel olarak art\u0131ml\u0131 olarak olu\u015fturulabilecek potansiyel \u00e7a\u011fr\u0131\u015f\u0131mlar ba\u011flam\u0131nda kelime \u00f6\u011frenimini sim\u00fcle eden yeniden \u00fcretilebilir ve \u00f6l\u00e7eklenebilir bir modeldir \u00f6zellikle art\u0131r\u0131ml\u0131 gizli Dirichlet tahsisi (Blei, Ng ve Jordan, 2003) konu modelleri. MKY indisleri, kelime s\u0131kl\u0131\u011f\u0131 ve entropi tahminlerinin yan\u0131 s\u0131ra edinim ya\u015f\u0131 ve s\u00f6zc\u00fcksel cevap gecikmelerinin insan puanlar\u0131 ile g\u00fc\u00e7l\u00fc ili\u015fkiler (kelime olgunlu\u011funun raporlanan performans\u0131n\u0131 a\u015fmaktad\u0131r) sa\u011flar.<\/p>\n\n

Do\u011fal Dil \u0130\u015fleme ve \u00d6\u011frenme Algoritmalar\u0131<\/h3>\n

DD\u0130, s\u00f6zc\u00fck say\u0131s\u0131, n-gram ve paragraf gibi basit tan\u0131mlay\u0131c\u0131 istatistiklerden1<\/sup><\/a> s\u00f6zc\u00fck, c\u00fcmle ve metin \u00f6zelliklerine kadar \u00e7ok say\u0131da dil y\u00fcz\u00fcn\u00fc tan\u0131mlamak i\u00e7in kullan\u0131labilir (Crossley, Allen, Kyle ve McNamara, 2014). \u015eekil 8.1'de g\u00f6sterildi\u011fi gibi, dilin bir\u00e7ok \u00f6zelli\u011fi s\u00f6zc\u00fcklerden toplan\u0131r (n-gram ve s\u00f6zc\u00fck torbalar\u0131 d\u00e2hil) ve hem g\u00f6zlemlenebilir \u00f6zellikleri analiz ederek (\u00f6r. kelime s\u0131kl\u0131klar\u0131, kelime-belge da\u011f\u0131l\u0131mlar\u0131) ve hem de metindeki gizli anlam\u0131 kullanarak elde edilebilir. (McNamara, 2011). Bilgi, kelimelerin \u00f6zellikleri, c\u00fcmleler ve metnin b\u00fct\u00fcn\u00fcyle sa\u011flan\u0131r. Bu bilgiler do\u011frusal regresyon, ay\u0131r\u0131c\u01312<\/sup><\/a> fonksiyon s\u0131n\u0131fland\u0131r\u0131c\u0131lar\u0131, Naif Bayes s\u0131n\u0131fland\u0131r\u0131c\u0131lar\u0131, destek vekt\u00f6r makineleri, yap\u0131sal ba\u011f\u0131nt\u0131 s\u0131n\u0131fland\u0131r\u0131c\u0131lar\u0131 ve karar a\u011fac\u0131 s\u0131n\u0131fland\u0131r\u0131c\u0131lar\u0131 gibi makine \u00f6\u011frenme teknikleri kullan\u0131larak analiz edilebilir. Bu teknikler \u00f6\u011frenme \u00e7\u0131kt\u0131lar\u0131n\u0131 tahmin etmek i\u00e7in kullan\u0131ld\u0131\u011f\u0131nda, daha sonra e\u011fitim teknolojileri veya uygulamalar\u0131nda kullan\u0131labilecek algoritmalar t\u00fcretilebilir. Bu uygulamalar\u0131n bir k\u0131sm\u0131n\u0131 a\u015fa\u011f\u0131daki b\u00f6l\u00fcmlerde tart\u0131\u015f\u0131yoruz.<\/p>\n

\"\"<\/p>\n<\/a> \u015eekil 8.1. DD\u0130 kullanarak algoritmalar geli\u015ftirmek, kelimeler, c\u00fcmleler ve metnin tamam\u0131 d\u00e2hil olmak \u00fczere, metin \u00fczerindeki \u00e7e\u015fitli bilgi kaynaklar\u0131na uygulanan makine \u00f6\u011frenme tekniklerini gerektirir.<\/span><\/i><\/span>\n

YAZININ DE\u011eERLEND\u0130R\u0130LMES\u0130<\/h2>\n

DD\u0130'nin e\u011fitim alan\u0131nda kullan\u0131lmas\u0131n\u0131n en yayg\u0131n \u00f6rne\u011fi, otomatik kompozisyon puanlama (OKP) algoritmalar\u0131n\u0131n geli\u015ftirilmesidir (Allen, Jacovina ve McNamara, 2016; Dikli, 2006; Weigle, 2013; Xi, 2010). OKP sistemleri \u00e7e\u015fitli yakla\u015f\u0131mlar\u0131 kullanarak kompozisyonlar\u0131 de\u011ferlendirir. \u00d6rne\u011fin Ak\u0131ll\u0131 Kompozisyon De\u011ferlendirici (Intelligent Essay Assessor-Landauer, Laham ve Foltz, 2003) bir metnin ba\u015fka bir \u00f6l\u00e7\u00fct metne benzerli\u011fini \u00f6ncelikle \u00d6SA ya dayanarak de\u011ferlendirir. Buna kar\u015f\u0131l\u0131k, Educational Testing Service'te geli\u015ftirilen e-rater (Burstein, Chodorow ve Leacock, 2004), Vantage Learning (Rudner, Garcia ve Welch, 2006) taraf\u0131ndan geli\u015ftirilen IntelliMetric Kompozisyon Puanlama Sistemi ve Wrting Pal gibi sistemler (McNamara, Crossley ve Roscoe, 2013) DD\u0130 teknikleri ve yapay zek\u00e2n\u0131n birle\u015fimine dayan\u0131r. OKP sistemleri kompozisyon gibi yazma \u00f6rneklerini i\u015fler ve yaz\u0131n\u0131n kalitesini ve i\u00e7eri\u011fe g\u00f6re do\u011frulu\u011funu de\u011ferlendirerek yazar\u0131n g\u00f6revin taleplerini ne derece yerine getirdi\u011fini de\u011ferlendirir. OKP teknolojileri olduk\u00e7a ba\u015far\u0131l\u0131d\u0131r, genel olarak uzman insan puanlay\u0131c\u0131lar kadar kesin do\u011fruluk seviyeleri rapor eder (Attali ve Burstein, 2006; Shermis, Burstein, Higgins ve Zechner, 2010; Valenti, Neri ve Cucchiarelli, 2003; Crossley, Kyle ve McNamara), 2015).<\/p>\n\n

Ak\u0131ll\u0131 \u00d6\u011fretici Sistemler<\/h3>\n

DD\u0130'nin bir di\u011fer kullan\u0131m\u0131, otomatik, ak\u0131ll\u0131 ders verme teknolojileri ba\u011flam\u0131nda olmu\u015ftur. DD\u0130, \u00f6zellikle ak\u0131ll\u0131 \u00f6\u011freticilerle diyalog yoluyla etkile\u015fimde bulunanlar (\u00f6r. AutoTutor: Graesser vd., 2004) ve \u00f6\u011frenenin s\u00f6zl\u00fc cevaplar vermesini isteyen (\u00f6r. AOD\u0130ES\u00d6) bir dizi ak\u0131ll\u0131 \u00f6\u011fretici sistemine (A\u00d6S) d\u00e2hil edilmi\u015ftir. (\u00f6r. AOD\u0130ES\u00d6: McNamara, Levinstein ve Boonthum, 2004; Writing Pal: McNamara vd., 2012; Roscoe ve McNamara, 2013). Bir \u00f6\u011frenen do\u011fal dil giri\u015fi yapt\u0131\u011f\u0131nda ve yararl\u0131 geri bildirim veya makul bir cevap bekledi\u011finde, DD\u0130 bu giri\u015fi yorumlamak ve uygun geri bildirim sa\u011flamak i\u00e7in kullan\u0131labilir (McNamara vd., 2013). Do\u011fal dili girdi olarak kabul eden \u00f6zel ders sistemleri i\u00e7in (\u00f6r. metin, problemler veya bilimsel s\u00fcre\u00e7lerin s\u00f6zl\u00fc a\u00e7\u0131klamalar\u0131), \u00f6\u011frenen cevaplar\u0131 a\u00e7\u0131k u\u00e7lu ve potansiyel olarak belirsiz olabilir. \u00d6rne\u011fin, \u00f6\u011frenciye, h\u00fccre mitozunun hangi faz\u0131n\u0131n mikrot\u00fcb\u00fcllerin uzat\u0131lmas\u0131n\u0131 i\u00e7erdi\u011fi sorulabilir. Bu t\u00fcr bir soru (\u00f6r. ne veya ne zaman sorusu) k\u0131sa cevaplar veya \u00e7oktan se\u00e7meli cevaplar kullan\u0131larak cevaplanabilir ve DD\u0130 gerektirmez. Buna kar\u015f\u0131l\u0131k, anafaz s\u00fcreci a\u00e7\u0131klamak i\u00e7in bir soru yayg\u0131n olarak \u00f6\u011frenciler aras\u0131nda farkl\u0131 muhtemel cevaplar temin eder. Bu nedenle, \u00f6\u011frenenin cevab\u0131n\u0131n do\u011frulu\u011funu ve kalitesini otomatik olarak tespit etmek DD\u0130 kullan\u0131m\u0131n\u0131 gerektirir.<\/p>\n

Neden sadece \u00e7oktan se\u00e7meli kullanm\u0131yorsunuz? Bir\u00e7ok \u00f6\u011fretici sistemi tam da bunu yap\u0131yor. Bununla birlikte, \u00f6\u011frencilerin nas\u0131l ve ni\u00e7in sorular\u0131na cevap vererek bir yap\u0131 veya olgu hakk\u0131nda derinlemesine bir anlay\u0131\u015f olu\u015fturma olas\u0131l\u0131klar\u0131 daha y\u00fcksektir (\u00f6r. Johnson-Glenberg, 2007; McKeown, Beck ve Blake, 2009; Wong, 1985). Ayr\u0131ca, \u00f6\u011frencilerin bu t\u00fcr sorulara verdikleri cevaplar\u0131n onlar\u0131n anlay\u0131\u015flar\u0131n\u0131n derinli\u011fini ortaya \u00e7\u0131karmas\u0131 muhtemeldir. (Graesser ve Person, 1994; Graesser, McNamara ve VanLehn, 2005; McNamara ve Kintsch, 1996). AutoTutor, zorlu konulara dair (\u00f6r. fizik, biyoloji, bilgisayar programlamas\u0131) \u00f6\u011frencilere derinlemesine nas\u0131l ve neden soru sormalar\u0131 gerekti\u011fini s\u00f6yleyerek e\u011fitim vermeye odaklanan bir A\u00d6S'dir. AutoTutor, \u00f6\u011freneni, do\u011fru cevaplar olu\u015fturmaya y\u00f6nlendiren bir diyalogda animasyonlu bir arac\u0131yla me\u015fgul eder. Bunu, ipucu, bilgi istemleri, iddialar, d\u00fczeltmeler ve \u00f6\u011frenen sorular\u0131na cevaplar gibi \u00e7e\u015fitli diyalog hareketlerini kullanarak yapar. Bu hamleler DD\u0130 tekniklerinin bir kombinasyonu ile ger\u00e7ekle\u015ftirilir. \u00d6rne\u011fin, AutoTutor, \u00f6\u011frencilerin belirli durumlarda \u00fcretebilecekleri c\u00fcmleleri (\u00f6r. bilmiyorum; anlamad\u0131m) ve do\u011fru cevab\u0131n \u00f6nemli k\u0131s\u0131mlar\u0131n\u0131 tespit etmek i\u00e7in de\u011fi\u015fmez ifadeler kullan\u0131r. AutoTutor ayr\u0131ca, \u00f6\u011frenenin verdi\u011fi cevap ile ideal cevap aras\u0131ndaki benzerli\u011fi tespit etmek i\u00e7in \u00d6SA'y\u0131 kullan\u0131r. Sabit ifadelerin, d\u00fczenli ifadelerin veya \u00f6r\u00fcnt\u00fclerin, \u00d6SA ile \u00f6\u011frencilerin s\u00f6zel cevaplar\u0131 ve beklentileri aras\u0131ndaki ters a\u011f\u0131rl\u0131kl\u0131 s\u0131kl\u0131k s\u00f6zc\u00fck \u00f6rt\u00fc\u015fmeleri, AutoTutor'un \u00f6\u011frenenin cevab\u0131n\u0131 anlaman\u0131n benzerini \u00fcretmeye izin verir ve bu benzetilmi\u015f anlay\u0131\u015f, \u00f6\u011frenci i\u00e7in uygun bir cevap \u00fcretir. (Graesser, bas\u0131m a\u015famas\u0131nda).<\/p>\n

AOD\u0130ES\u00d6 (Aktif Okuma ve D\u00fc\u015f\u00fcnmeye Y\u00f6nelik Etkile\u015fimli Strateji E\u011fitimi), a\u00e7\u0131k u\u00e7lu cevaplara cevap vermek i\u00e7in DD\u0130 tekniklerinin birle\u015fimine dayanan bir di\u011fer A\u00d6S'dir. AOD\u0130ES\u00d6 hem DD\u0130 hem de bilgi i\u015flemsel dilbilim literat\u00fcr\u00fcnde zorlu bir g\u00f6rev olan \u00f6\u011frenenin kendi a\u00e7\u0131klamalar\u0131ndaki yorumlama sorununu ele alan ilk otomatik sistemler aras\u0131ndayd\u0131. AOD\u0130ES\u00d6, \u00f6\u011frencilerin kendini a\u00e7\u0131klamay\u0131 (\u00f6r. metni kendi kendine a\u00e7\u0131klama s\u00fcreci), ili\u015fkilendirici ve ayr\u0131nt\u0131l\u0131 \u00e7\u0131kar\u0131mlar \u00fcretme gibi anlama stratejileriyle birlikte kullanma amac\u0131yla \u00f6\u011fretim ve pratik sa\u011flayarak zorlu bilimsel metinleri kavray\u0131\u015flar\u0131n\u0131 geli\u015ftirir. AOD\u0130ES\u00d6 \u00f6\u011fretiminin uygulama a\u015famas\u0131nda, \u00f6\u011frenciler zorlu metinler i\u00e7in kendi a\u00e7\u0131klamalar\u0131n\u0131 \u00fcretirler. \u00d6\u011frencilerin AOD\u0130ES\u00d6'taki kendi a\u00e7\u0131klamalar\u0131, kelimeler hakk\u0131ndaki g\u00f6zlemlenebilir ve gizli anlamsal bilgilerin bir birle\u015fimini kullanarak kendi a\u00e7\u0131klamalar\u0131ndaki ve metindeki s\u00f6zc\u00fcklerden gelen bilgileri birle\u015ftiren bir algoritma kullan\u0131larak puanlan\u0131r, (McNamara, Boonthum, Levinstein ve Millis, 2007) ). Algoritma otomatik olarak her bir ki\u015fisel a\u00e7\u0131klama i\u00e7in 0 ile 3 aras\u0131nda bir puan atar. Daha y\u00fcksek puanlar, metin i\u00e7eri\u011fi (hem hedef c\u00fcmle hem de daha \u00f6nce okunan c\u00fcmleler) ile ilgili bilgileri i\u00e7eren ki\u015fisel a\u00e7\u0131klamalara, d\u00fc\u015f\u00fck puanlar ise ilgisiz veya k\u0131sa cevaplara verilir. Puanlama algoritmas\u0131, \u00f6\u011frencilerin hedef c\u00fcmle, \u00f6nceki metin i\u00e7eri\u011fi ve d\u00fcnya bilgisi aras\u0131nda ne \u00f6l\u00e7\u00fcde ba\u011flant\u0131 kurduklar\u0131n\u0131 yans\u0131tacak \u015fekilde tasarlanm\u0131\u015ft\u0131r. Sistem \u00e7ok \u00e7e\u015fitli metinlerdeki a\u00e7\u0131klamalar\u0131n insanlar taraf\u0131ndan geli\u015ftirilen puanlar\u0131n\u0131 ba\u015far\u0131yla e\u015fle\u015ftirir (Jackson, Guess ve McNamara, 2010; McNamara vd., 2007).<\/p>\n\n

Bilgisayar Destekli \u0130\u015fbirlikli \u00d6\u011frenme (BD\u0130\u00d6)<\/h3>\n

DD\u0130 teknikleri, i\u015fbirlikli \u00f6\u011frenme ortamlar\u0131nda ve \u00f6zellikle Bilgisayar Destekli \u0130\u015fbirlikli \u00d6\u011frenme (BD\u0130\u00d6) sistemlerinde ortaya \u00e7\u0131kan s\u00f6ylemde de uygulanm\u0131\u015ft\u0131r (Stahl, 2006). Bu sistemlerin bir alt k\u00fcmesi, Bakhtin (1981) taraf\u0131ndan ortaya konan, daha sonra BD\u0130\u00d6 i\u00e7in bir paradigma olarak ortaya \u00e7\u0131km\u0131\u015f olan bir kavram olan diyalojiye dayal\u0131 BD\u0130\u00d6 konu\u015fmalar\u0131n\u0131 modellemektedir (Koschmann, 1999). Dong'un tak\u0131m ileti\u015fimi tasar\u0131m\u0131, Polyphony (Trausan-Matu, Rebedea, Dragan ve Alexandru, 2007), Bilgi Alan\u0131 G\u00f6r\u00fcnt\u00fcleyici (Teplovs, 2008) ve Reader Bench'tir (Dascalu, Stavarache vd., 2015; Dascalu, Trausan-Matu, McNamara ve Dessus, 2015) en temsili yakla\u015f\u0131mlard\u0131r. ReaderBench, i\u015f birli\u011fine dayal\u0131 \u00f6\u011frenmenin yan\u0131 s\u0131ra dili anlama ile ilgili birden fazla hedefe ula\u015fmak i\u00e7in metin madencili\u011fi teknikleri, geli\u015fmi\u015f DD\u0130 ve sosyal a\u011f analizinin g\u00fcc\u00fcnden yararlanmaktad\u0131r (Dascalu, 2014). ReaderBench, kat\u0131l\u0131mc\u0131lar aras\u0131nda iletilen bilgilerin anlamsal metinsel b\u00fct\u00fcnle\u015fme yoluyla hesapland\u0131\u011f\u0131 bir Uyum A\u011f\u0131 Analizi perspektifi ile kat\u0131l\u0131m ve i\u015f birli\u011fini modellemektedir (Dascalu, Trausan-Matu, Dessus ve McNamara, 2015a). Ayr\u0131ca ReaderBench, polifonik s\u00f6ylem modeline dayanan i\u015f birli\u011fini de\u011ferlendirmek i\u00e7in otomatik bir diyalog modeli ortaya koymu\u015ftur (Trausan-Matu, Stahl ve Sarmiento, 2007). Diyaloji kuramlar\u0131na dayanarak (Bakhtin, 1981), sistem otomatik olarak sesleri veya kat\u0131l\u0131mc\u0131n\u0131n g\u00f6r\u00fc\u015f\u00fcn\u00fc, t\u00fcm konu\u015fmay\u0131 kapsayan s\u0131k\u0131 bir \u015fekilde birbirine ba\u011fl\u0131 veya anlamsal olarak ilgili kavramlar\u0131 i\u00e7eren anlamsal zincirler olarak tan\u0131mlar (Dascalu, Trausan-Matu, Dessus ve McNamara, 2015b). Bu nedenle, i\u015f birli\u011fi, farkl\u0131 kat\u0131l\u0131mc\u0131lar aras\u0131nda fikir al\u0131\u015fveri\u015fini vurgulamak i\u00e7in kullan\u0131lan ortak-olu\u015f \u00f6r\u00fcnt\u00fclerinde bilgi i\u015flemsel olarak yakalanan farkl\u0131 kat\u0131l\u0131mc\u0131 seslerinin kar\u015f\u0131l\u0131kl\u0131 canland\u0131rmas\u0131ndan ortaya \u00e7\u0131kar.<\/p>\n\n

Kitlesel A\u00e7\u0131k \u00c7evrimi\u00e7i Dersler (KA\u00c7D'ler)<\/h3>\n

DD\u0130'nin bir di\u011fer kullan\u0131m\u0131, \u00e7evrimi\u00e7i dersler, \u00f6zellikle de b\u00fcy\u00fck a\u00e7\u0131k \u00e7evrimi\u00e7i dersler (KA\u00c7D'ler) ba\u011flam\u0131nda olmu\u015ftur. KA\u00c7D'ler, binlerce \u00f6\u011frenciye dersleri \u00fccretsiz olarak sunmak i\u00e7in \u00e7evrimi\u00e7i platformlar\u0131 kullan\u0131r. KA\u00c7D'ler, uzaktan ve ya\u015fam boyu \u00f6\u011frenenlere eri\u015filebilirli\u011fi artt\u0131rma potansiyelleri nedeniyle \u00f6vg\u00fcyle kar\u015f\u0131lanmaktad\u0131r (Koller, Ng, Do ve Chen, 2013). Bu platformlar, tart\u0131\u015fma ak\u0131\u015flar\u0131nda ve e-postalarda \u00f6\u011frencilerin olu\u015fturdu\u011fu dillerin yan\u0131 s\u0131ra t\u0131klama ak\u0131\u015f\u0131 g\u00fcnl\u00fckleri, \u00f6devler, kurs performans\u0131 ve \u00e7ok b\u00fcy\u00fck miktarda veri sa\u011flayabilir. Bu veriler \u00f6\u011frenen tutumlar\u0131, tamamlama ve \u00f6\u011frenmeyi incelemek i\u00e7in ara\u015ft\u0131r\u0131labilir (Seaton, Bergner, Chuang, Mitros ve Pritchard, 2014; Wen, Yang ve Rose, 2014a, 2014b).<\/p>\n

KA\u00c7D'lerde \u00f6\u011frenen dilini analiz etmek i\u00e7in en yayg\u0131n DD\u0130 yakla\u015f\u0131m\u0131, duygular\u0131 analiz eden ara\u00e7lardan olmu\u015ftur. Duygu analizi, olumlu ya da olumsuz duygular\u0131n dilini ya da motivasyon, anla\u015fma, bili\u015fsel mekanizmalar ya da kat\u0131l\u0131mla ilgili kelimeleri inceler (Chaturvedi, Goldwasser ve Daume, 2014; Elouazizi, 2014; Moon, Potdar ve Martin, 2014; Wen vd., 2014a, 2014b). \u00d6rne\u011fin, Moon vd. (2014), \u00f6\u011frenen liderlerini tan\u0131mlamak i\u00e7in kat\u0131l\u0131mc\u0131lar aras\u0131nda duygu terimleri ve anlamsal benzerlikler kullanm\u0131\u015ft\u0131r. Bak\u0131\u015f a\u00e7\u0131s\u0131na ili\u015fkin dil indislerinin (\u00f6r. san\u0131r\u0131m, bence b\u00fcy\u00fck ihtimalle, muhtemelen) kursun d\u00fc\u015f\u00fck kat\u0131l\u0131m d\u00fczeyleriyle ili\u015fkili oldu\u011funu g\u00f6stermi\u015ftir. Wen ve meslekta\u015flar\u0131, (2014a, 2014b), \u00f6\u011frencilerin tart\u0131\u015fma zamirlerini ve tart\u0131\u015fma forumlar\u0131ndaki motivasyonla ilgili s\u00f6zc\u00fckleri kullanmalar\u0131n\u0131n, dersten ayr\u0131lma riskinin daha d\u00fc\u015f\u00fck olaca\u011f\u0131n\u0131 \u00f6ng\u00f6rd\u00fc\u011f\u00fcn\u00fc bulmu\u015flard\u0131r.<\/p>\n

Benzer \u015fekilde, Crossley, McNamara vd. (2015), \u00f6\u011frencilerin dilini, KA\u00c7D tart\u0131\u015fma forumunda, e\u011fitsel veri madencili\u011fi konusunu kapsayan bir kursta incelemek i\u00e7in \u00e7ok say\u0131da dil \u00f6zelli\u011fi kullanm\u0131\u015ft\u0131r (Baker vd., bas\u0131m a\u015famas\u0131nda). Crossley, McNamara vd. (2015) KA\u00c7D tart\u0131\u015fma forumlar\u0131nda kat\u0131lan 320 \u00f6\u011frencinin (\u00f6r. g\u00f6nderilen 49 kelime) tamamlama oranlar\u0131n\u0131 ba\u015far\u0131 ile (%70 do\u011frulukla) tahmin etmi\u015ftir. Kursta bitirme sertifikas\u0131 alma olas\u0131l\u0131\u011f\u0131 daha y\u00fcksek olan \u00f6\u011frenciler genellikle daha karma\u015f\u0131k bir dil kulland\u0131lar. \u00d6rne\u011fin, onlar\u0131n mesajlar\u0131 daha anla\u015f\u0131l\u0131r ve tutarl\u0131, daha s\u0131k ve belirli bir kelime kullan\u0131lm\u0131\u015f ve daha genel yazma niteli\u011fine sahipti. \u0130lgin\u00e7tir ki, duyu\u015f ile ilgili indisler tamamlanma oranlar\u0131n\u0131 \u00f6ng\u00f6rm\u00fcyordu.<\/p>\n

Toplu olarak, bu ara\u015ft\u0131rma DD\u0130'nin KA\u00c7D'lerin \u00f6\u011fretim g\u00f6revlisi ile \u00f6\u011frenciler aras\u0131ndaki ve \u00f6\u011frencilerin kendi aras\u0131ndaki ileti\u015fimi ba\u011flam\u0131nda g\u00fc\u00e7l\u00fc bir ba\u015far\u0131 g\u00f6stergesi olabilece\u011fine dair umut verici kan\u0131tlar sunar ve bu \u00f6zellikle de uzaktan kurslar i\u00e7in \u00e7ok \u00f6nemlidir. Ayr\u0131ca, bu ileti\u015fim daha sonra \u00f6\u011frenen performans\u0131n\u0131n de\u011ferlendirme formlar\u0131 olarak da kullan\u0131labilir. Bu nedenle, KA\u00c7D'lerin \u00f6\u011frenen kat\u0131l\u0131m\u0131n\u0131 ve potansiyel ba\u015far\u0131s\u0131n\u0131 daha iyi izlemek i\u00e7in tart\u0131\u015fma forumlar\u0131n\u0131 i\u00e7ermesi gerekti\u011fi a\u00e7\u0131k g\u00f6r\u00fcnmektedir. \u00d6\u011frencilerin kulland\u0131\u011f\u0131 dil, kursu tamamlama olas\u0131l\u0131\u011f\u0131 daha d\u00fc\u015f\u00fck olan \u00f6\u011frencileri belirlemek ve bu \u00f6\u011frencilere e-posta g\u00f6ndermek, i\u00e7erik \u00f6nermek veya \u00f6zel ders \u00f6nermek gibi m\u00fcdahaleler hedeflemek i\u00e7in de kullan\u0131labilir. Dil anlay\u0131\u015f\u0131n\u0131 otomatikle\u015ftirmek ve b\u00f6ylece bu kurslardaki dil ve sosyal etkile\u015fimler hakk\u0131nda bilgi vermek, KA\u00c7D'lerde hem \u00f6\u011frenmeyi hem de etkile\u015fimi geli\u015ftirmeye yard\u0131mc\u0131 olacakt\u0131r.<\/p>\n\n

DD\u0130'nin G\u00dcC\u00dc<\/h2>\n

DD\u0130, \u00f6ncelikle dilin her yerde olmas\u0131 ve ayn\u0131 zamanda dili analiz etme ara\u00e7lar\u0131n\u0131n dilin neredeyse her y\u00f6n\u00fcyle ilgili g\u00f6stergeler sa\u011flamas\u0131 nedeniyle olduk\u00e7a g\u00fc\u00e7l\u00fcd\u00fcr (Crossley, 2013). DD\u0130 kullan\u0131lan belirli kelimeleri, kelime gruplar\u0131n\u0131 ve kelimeler aras\u0131ndaki ve daha b\u00fcy\u00fck metin g\u00f6vdeleri aras\u0131ndaki ili\u015fkilerin g\u00fcc\u00fcn\u00fc tespit edebilir. Ayr\u0131ca, metnin s\u0131kl\u0131\u011f\u0131, somutlu\u011fu veya anlaml\u0131l\u0131\u011f\u0131, c\u00fcmlelerin karma\u015f\u0131kl\u0131\u011f\u0131 ve metnin uyum ve t\u00fcr gibi \u00e7e\u015fitli y\u00f6nleri gibi metnin \u00f6zelliklerini de alg\u0131layabilir. Kelimeler ve \u00f6zellikleri, \u00e7e\u015fitli yap\u0131lar i\u00e7in vekil g\u00f6revi g\u00f6r\u00fcr. \u00d6rne\u011fin, bir metindeki kelimelerin s\u0131kl\u0131\u011f\u0131, metni anlamak i\u00e7in gerekli olabilecek bilgiyi tahmin etmede bir vekil olarak hizmet eder. Bir metnin birle\u015ftirilmesi, bir metindeki bo\u015fluklar\u0131 doldurmak i\u00e7in gerekli olan bilginin bir tahminini verir.<\/p>\n

DD\u0130, \u00e7ok \u00e7e\u015fitli ba\u015fka yap\u0131lar\u0131 tan\u0131mlamak i\u00e7in kullan\u0131lm\u0131\u015ft\u0131r. \u00d6rne\u011fin, Crossley ve McNamara (2012), ikinci dil (D2) yazarlar\u0131n\u0131n makalelerinin dil \u00f6zelliklerinin, bu yazarlar\u0131n ana dilini tahmin edebilece\u011fini g\u00f6stermi\u015ftir. Varner, Roscoe ve McNamara (2013) hem Coh-Metrix hem de DSKS taraf\u0131ndan sa\u011flanan g\u00f6stergeleri kullanarak \u00f6\u011frencilerin ve \u00f6\u011fretmenlerin kompozisyon kalitesi puanlar\u0131ndaki farkl\u0131l\u0131klar\u0131 incelemi\u015ftir. Louwerse, McCarthy, McNamara ve Graesser (2004), konu\u015fulan veyaz\u0131l\u0131 \u0130ngilizce \u00f6rnekleri aras\u0131ndaki farklar\u0131 belirlemek i\u00e7in DD\u0130 tekniklerini kulland\u0131. McCarthy, Briner, Rus ve McNamara (2007) Coh-Metrix'in giri\u015f, y\u00f6ntem, sonu\u00e7 ve tart\u0131\u015fma gibi tipik bilimsel metinlerdeki b\u00f6l\u00fcmleri farkl\u0131la\u015ft\u0131rabildi\u011fini g\u00f6stermi\u015ftir. Ek olarak, Crossley, Louwerse, McCarthy ve McNamara'n\u0131n (2007) ikinci dil \u00f6\u011frenenlerin metinlerinin incelemeleri, ikinci dil \u00f6\u011frenme ama\u00e7lar\u0131 i\u00e7in edinilmi\u015f (veya otantik) ile adapte edilmi\u015f (veya basitle\u015ftirilmi\u015f) metinler aras\u0131nda \u00e7ok \u00e7e\u015fitli yap\u0131sal ve s\u00f6zc\u00fcksel farkl\u0131l\u0131klar ortaya koydu. Son olarak, DD\u0130 aldat\u0131c\u0131l\u0131\u011f\u0131 tespit etmek i\u00e7in de kullan\u0131lm\u0131\u015ft\u0131r. Duran, Hall, McCarthy ve McNamara (2010), bir ki\u015finin aldat\u0131c\u0131 oldu\u011fu konu\u015fma diyaloglar\u0131 ile samimi oldu\u011fu konu\u015fmalar aras\u0131nda dilin hangi \u00f6zelliklerinin ay\u0131r\u0131c\u0131 oldu\u011funu ara\u015ft\u0131rd\u0131.<\/p>\n

DD\u0130'nin kullanman\u0131n olas\u0131 sak\u0131ncalar\u0131 oldu\u011funa dikkat etmek \u00f6nemlidir. \u00d6rne\u011fin, belirli DD\u0130 teknikleri, kelime say\u0131mlar\u0131n\u0131 veya \"s\u00f6zc\u00fck \u00e7antas\u0131\" yakla\u015f\u0131mlar\u0131n\u0131 kullanan diyalogun basitle\u015ftirilmi\u015f temsillerine dayan\u0131r. \u00d6SA vekt\u00f6r uzaylar\u0131, gizli Dirichlet tahsisi konu da\u011f\u0131l\u0131mlar\u0131 (GDT; Blei vd., 2003) ve sinir a\u011flar\u0131na dayal\u0131 word2vec modelleri d\u00e2hil olmak \u00fczere en \u00f6nemli ve en yayg\u0131n kullan\u0131lan DD\u0130 kelime g\u00f6sterimlerinin (Mikolov, Chen, Corrado ve Dean, 2013) hepsi, kelime s\u0131ras\u0131n\u0131n dikkate al\u0131nmad\u0131\u011f\u0131 \u201cs\u00f6zc\u00fck \u00e7antas\u0131\u201d varsay\u0131m\u0131na tabidir. Ek olarak, bir\u00e7ok DD\u0130 analizi, konu\u015fmac\u0131n\u0131n niyetleri veya pragmatik y\u00f6nleri gibi ba\u011flamlar\u0131 g\u00f6rmezden gelir. Benzer \u015fekilde, DD\u0130 analizleri genellikle belirli \u015firketlerle ve durumlarla s\u0131n\u0131rl\u0131d\u0131r ve di\u011fer ba\u011flamlara genellenememektedir. Bu (ve di\u011fer) uyar\u0131larla bile DD\u0130 son derece g\u00fc\u00e7l\u00fcd\u00fcr. DD\u0130 ara\u00e7lar\u0131ndan edinilebilecek geni\u015f bilgi kaynaklar\u0131 nedeniyle ve kulland\u0131\u011f\u0131m\u0131z dil, d\u00fc\u015f\u00fcnceler ve niyetleri temsil eden bir uzant\u0131 veya haricile\u015ftirme olabilece\u011finden DD\u0130 bireyler, yetenekleri, duygular\u0131, niyetleri ve sosyal etkile\u015fimleri hakk\u0131nda bilgi sa\u011flayabilir. \u00d6\u011frenme analiti\u011fi ba\u011flam\u0131nda, \u00f6\u011frenme s\u00fcre\u00e7lerini ve \u00f6\u011freneni otomatik olarak anlama yolunda bir ara\u00e7t\u0131r.<\/p>\n\n

B\u00fcy\u00fck Resim<\/h3>\n

DD\u0130, ara\u015ft\u0131rmac\u0131lar\u0131n dil ve onun ya\u015fam\u0131n \u00e7e\u015fitli y\u00f6nlerinde potansiyel olarak oynad\u0131\u011f\u0131 rolleri daha iyi anlayabilmelerini sa\u011flayan dil analizini otomatikle\u015ftiren teknikler sunar. DD\u0130, \u00f6\u011freneni sorulara, a\u00e7\u0131klamalara ve kompozisyon cevaplar\u0131 i\u00e7inde bir dil olu\u015fturmas\u0131 i\u00e7in y\u00f6nlendiren ak\u0131ll\u0131 \u00f6\u011fretici sistemlerindeki geri bildirim sistemlerini bilgilendirir. DD\u0130 ders tabanl\u0131 sistemler dil i\u00e7inde ak\u0131ll\u0131 sim\u00fclasyonu i\u00e7in bir ara\u00e7 sa\u011flar. DD\u0130 ayr\u0131ca \u00e7evrimi\u00e7i tart\u0131\u015fma forumlar\u0131 ba\u011flam\u0131nda da bilgilendiricidir. \u00d6\u011frencilerin iyi performans g\u00f6sterme veya kursu tamamlama ihtimalini yordayarak, \u00f6\u011frenci tutumlar\u0131, motivasyonu ve dilin kalitesi hakk\u0131nda bilgi sa\u011flar.<\/p>\n

\u00d6\u011frenme analiti\u011finin bir amac\u0131, daha etkili bir \u00f6\u011fretim sa\u011flamak i\u00e7in \u00f6\u011frencilerin \u00f6zelliklerini ve becerilerini modellemektir (Allen ve McNamara, 2015). Bu verileri \u00f6zellikle, \u00e7e\u015fitli ama\u00e7lar i\u00e7in kullanabiliriz: Performansla ilgili otomatik geri bildirim sa\u011flama, \u00f6\u011frenme s\u0131ras\u0131nda m\u00fcdahale etme, y\u00f6nlendirme veya bili\u015fsel deste\u011fi sa\u011flama, \u00f6zel ders \u00f6nerme, analiz verilerinden elde edilen bilgilerin \u00f6\u011frenimi geli\u015ftirece\u011fi varsay\u0131m\u0131yla izleme, \u00f6\u011frenmeyi ki\u015fiselle\u015ftirme vb. Bu ama\u00e7la, ara\u015ft\u0131rmac\u0131lar giderek daha b\u00fcy\u00fck, karma\u015f\u0131k veri kaynaklar\u0131na (yani, b\u00fcy\u00fck veriye) d\u00f6nmekte ve \u00e7e\u015fitli veri t\u00fcrleri ve analitik tekniklerin bile\u015fimini kullanmaktad\u0131r. Bu \u00e7aba i\u00e7in DD\u0130 \u00e7ok \u00f6nemlidir, \u00e7\u00fcnk\u00fc \u00f6nerilen teknikler, \u00e7e\u015fitli ba\u011flamlarda anlama d\u00fczeyini tahmin etme ve de\u011ferlendirme yoluyla \u00f6\u011frenenin \u00f6\u011frenmesini geli\u015ftirmeye yard\u0131mc\u0131 olur. Ancak DD\u0130 bulmacan\u0131n yaln\u0131zca bir par\u00e7as\u0131d\u0131r.<\/p>\n

\"\"<\/p>\n<\/a> \u015eekil 8.2. E\u011fitsel sonu\u00e7lar\u0131n\u0131n \u00f6ng\u00f6r\u00fclmesi, birden fazla veri kayna\u011f\u0131n\u0131n birle\u015ftirilmesini gerektirecektir.<\/i><\/span>\n

\u015eekil 8.2'de g\u00f6sterildi\u011fi gibi, \u00f6\u011frenen \u00e7\u0131kt\u0131lar\u0131n\u0131n eksiksiz ve son derece kestirimsel bir anlay\u0131\u015f\u0131n\u0131n geli\u015ftirilmesi, \u00e7ok say\u0131da bilgi kayna\u011f\u0131n\u0131 ve veri analizine y\u00f6nelik \u00e7e\u015fitli yakla\u015f\u0131mlar\u0131 gerektirir. \u00d6\u011frenme, \u00e7oklu katmanlara ve \u00e7oklu zaman \u00f6l\u00e7eklerine sahip karma\u015f\u0131k bir s\u00fcre\u00e7tir. \u00d6\u011frenme s\u00fcrecini anlamak i\u00e7in herhangi bir kayna\u011fa veya veri t\u00fcr\u00fcne g\u00fcvenmek, \u00f6zellikle \u015fu anda \u00e7ok say\u0131da otomatik bilgi kayna\u011f\u0131 mevcut oldu\u011funda, uza\u011f\u0131 g\u00f6remeyen bir yakla\u015f\u0131md\u0131r. DD\u0130, nihai olarak arad\u0131\u011f\u0131m\u0131z b\u00fcy\u00fck resmin ayr\u0131lmaz bir par\u00e7as\u0131 olarak giderek daha fazla tan\u0131nan bir veri kayna\u011f\u0131d\u0131r. Tam bir \u00f6\u011frenme anlay\u0131\u015f\u0131 geli\u015ftirmek, birden fazla veri kayna\u011f\u0131n\u0131n birle\u015ftirilmesini gerektirecektir.<\/p>\n\n

TE\u015eEKK\u00dcR B\u00d6L\u00dcM\u00dc<\/h2>\n

Bu b\u00f6l\u00fcm\u00fcn bir k\u0131sm\u0131 E\u011fitim Bilimleri Enstit\u00fcs\u00fc (EBE R305A120707, R305A130124), Ulusal Bilim Vakf\u0131 (UBV DRL-1319645, DRL-1418352, DRL-1418378, DRL-1417997) ve Denizcilik Ara\u015ft\u0131rma B\u00fcrosu taraf\u0131ndan desteklenmi\u015ftir. Ara\u015ft\u0131rma (DAB N000141410343). Bu yaz\u0131da dile getirilen herhangi bir g\u00f6r\u00fc\u015f, bulgu ve \u00e7\u0131kar\u0131mlar veya tavsiye yazarlara aittir ve IES, UBV veya DAB'nin g\u00f6r\u00fc\u015flerini yans\u0131tmak zorunda de\u011fildir. DD\u0130 konusundaki ara\u015ft\u0131rmalar\u0131m\u0131za y\u0131llar boyunca katk\u0131da bulunan bir\u00e7ok \u00f6\u011frenciye, doktora sonras\u0131 ara\u015ft\u0131rmac\u0131ya ve fak\u00fclteye minnettar\u0131z.<\/p>\n\n

KAYNAK\u00c7A<\/h2>\nAllen, L. K., Jacovina, M. E., & McNamara, D. S. (2016). Computer-based writing instruction. In C. A. MacArthur, S. Graham, & J. Fitzgerald (Eds.), <\/span>Handbook of writing research<\/i><\/span>, 2nd ed. (S. 316-329). New York: The Guilford Press. <\/span><\/span>\n\nAllen, L. K., & McNamara, D. S. (2015). You are your words: Modeling students' vocabulary knowledge with natural language processing. In O. C. Santos, J. G. Boticario, C. Romero, M. Pechenizkiy, A. Merceron, P. Mitros, J. M. Luna, C. Mihaescu, P. Moreno, A. Hershkovitz, S. Ventura, & M. Desmarais (Eds.), Proceedings of the 8th International Conference on Educational Data Mining <\/i>(EDM 2015) 26\u201329 June 2015, Madrid, Spain (pp. 258\u2013265). International Educational Data Mining Society. <\/span>\n\nAttali, Y., & Burstein, J. (2006). Automated essay scoring with e-rater\u00ae V. 2. The Journal of Technology, Learning and Assessment, 4<\/i>(2). doi:10.1002\/j.2333-8504.2004.tb01972.x <\/span>\n\nBaker, R., Wang, E., Paquette, L., Aleven, V., Popescu, O., Sewall, J., Rose, C., Tomar, G., Ferschke, O., Hollands, F., Zhang, J., Cennamo, M., Ogden, S., Condit, T., Diaz, J., Crossley, S., McNamara, D., Comer, D., Lynch, C., Brown, R., Barnes, T., & Bergner, Y. (in press). A MOOC on educational data mining. In. S. ElAtia, O. Za\u00efane, & D. Ipperciel (Eds.). Data Mining and Learning Analytics in Educational Research<\/i>. Wiley & Blackwell. <\/span>\n\nBakhtin, M.M. (1981). The dialogic imagination: Four essays <\/i>(C. Emerson & M. Holquist, Trans.). Austin, TX: University of Texas Press. <\/span>\n\nBlei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent Dirichlet allocation. Journal of Machine Learning Research, 3<\/i>(4\u20135), 993\u20131022. <\/span>\n\nBurstein, J., Chodorow, M., & Leacock, C. (2004). Automated essay evaluation: The Criterion online writing service. Ai Magazine, 25<\/i>(3), 27.<\/span>\n\nChaturvedi, S., Goldwasser, D., & Daum\u00e9 III, H. (2014). Predicting instructor's intervention in MOOC forums. In D. Marcu, K. Toutanova, & H. W. Baidu (Eds.), Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics <\/i>(pp. 1501\u20131511). Baltimore, MD. <\/span>\n\nCrossley, S. A. (2013). Advancing research in second language writing through computational tools and machine learning techniques: A research agenda. Language Teaching, 46<\/i>(2), 256\u2013271. <\/span>\n\nCrossley, S. A., Allen, L. K., Kyle, K., & McNamara, D. S. (2014). Analyzing discourse processing using a simple natural language processing tool (SiNLP). Discourse Processes, 51<\/i>, 511\u2013534. <\/span>\n\nCrossley, S. A., Kyle, K., & McNamara, D. S. (2015). To aggregate or not? Linguistic features in automatic essay scoring and feedback systems. Journal of Writing Assessment, 8<\/i>(1). http:\/\/www.journalofwritingassessment.org\/article.php?article=80 <\/span>\n\nCrossley, S. A. Kyle, K., & McNamara, D. S. (in press). Tool for the automatic analysis of text cohesion (TAACO): Automatic assessment of local, global, and text cohesion. Behavior Research Methods<\/i>. <\/span>\n\nCrossley, S. A., & Louwerse, M. (2007). Multi-dimensional register classification using bigrams. International Journal of Corpus Linguistics, 12<\/i>(4), 453\u2013478. <\/span>\n\nCrossley, S. A., Louwerse, M., McCarthy, P. M., & McNamara, D. S. (2007). A linguistic analysis of simplified and authentic texts. Modern Language Journal, 91<\/i>, 15\u201330. <\/span>\n\nCrossley, S. A., & McNamara, D. S. (2012). Interlanguage talk: A computational analysis of non-native speakers' lexical production and exposure. In P. M. McCarthy & C. Boonthum-Denecke (Eds.), Applied natural language processing and content analysis: Identification, investigation, and resolution <\/i>(pp. 425\u2013437). Hershey, PA: IGI Global. <\/span>\n\nCrossley, S. A., McNamara, D. S., Baker, R., Wang, Y., Paquette, L., Barnes, T., & Bergner, Y. (2015). Language to completion: Success in an educational data mining massive open online class. In O. C. Santos, J. G. Boticario, C. Romero, M. Pechenizkiy, A. Merceron, P. Mitros, J. M. Luna, C. Mihaescu, P. Moreno, A. Hershkovitz, S. Ventura, & M. Desmarais (Eds.), Proceedings of the 8th International Conference on Educational Data Mining <\/i>(EDM 2015), 26\u201329 June 2015, Madrid, Spain (pp. 388\u2013391). International Educational Data Mining Society. <\/span>\n\nDascalu, M. (2014). Analyzing discourse and text complexity for learning and collaborating. Studies in Computational Intelligence <\/i>(Vol. 534). Switzerland: Springer. <\/span>\n\nDascalu, M., Stavarache, L. L., Dessus, P., Trausan-Matu, S., McNamara, D. S., & Bianco, M. (2015). ReaderBench: The learning companion. In A. Mitrovic, F. Verdejo, C. Conati, & N. Heffernan (Eds.), Proceedings of the 17th International Conference on Artificial Intelligence in Education <\/i>(AIED\u201915), 22\u201326 June 2015, Madrid, Spain (pp. 915\u2013916). Springer. <\/span>\n\nDascalu, M., Trausan-Matu, S., Dessus, P., & McNamara, D. S. (2015a). Discourse cohesion: A signature of collaboration. In P. Blikstein, A. Merceron, & G. Siemens (Eds.), Proceedings of the 5th International Learning Analytics & Knowledge Conference <\/i>(LAK\u201915), 16\u201320 March, Poughkeepsie, NY, USA (pp. 350\u2013354). New York: ACM. <\/span>\n\nDascalu, M., Trausan-Matu, S., Dessus, P., & McNamara, D. S. (2015b). Dialogism: A framework for CSCL and a signature of collaboration. In O. Lindwall, P. H\u00e4kkinen, T. Koschmann, P. Tchounikine, & S. Ludvigsen (Eds.), Proceedings of the 11th International Conference on Computer-Supported Collaborative Learning <\/i>(CSCL 2015), 7\u201311 June 2015, Gothenburg, Sweden (pp. 86\u201393). International Society of the Learning Sciences. <\/span>\n\nDascalu, M., Trausan-Matu, S., McNamara, D. S., & Dessus, P. (2015). ReaderBench: Automated evaluation of collaboration based on cohesion and dialogism. International Journal of Computer-Supported Collaborative Learning, 10<\/i>(4), 395\u2013423. <\/span>\n\nDascalu, M., McNamara, D. S., Crossley, S. A., & Trausan-Matu, S. (2016). Age of exposure: A model of word learning. Proceedings of the 30th Conference on Artificial Intelligence <\/i>(AAAI-16), 12\u201317 February 2016, Phoenix, Arizona, USA (pp. 2928\u20132934). Palo Alto, CA: AAAI Press. <\/span>\n\nDikli, S. (2006). An overview of automated scoring of essays. The Journal of Technology, Learning and Assessment, 5<\/i>(1). http:\/\/files.eric.ed.gov\/fulltext\/EJ843855.pdf<\/span>\n\nDong, A. (2005). The latent semantic approach to studying design team communication. Design Studies, 26<\/i>(5), 445\u2013461. <\/span>\n\nDuran, N. D., Hall, C., McCarthy, P. M., & McNamara, D. S. (2010). The linguistic correlates of conversational deception: Comparing natural language processing technologies. Applied Psycholinguistics, 31<\/i>(3), 439\u2013462. <\/span>\n\nElouazizi, N. (2014). Point-of-view mining and cognitive presence in MOOCs: A (computational) linguistics perspective. EMNLP 2014<\/i>, 32. http:\/\/www.aclweb.org\/anthology\/W14-4105 <\/span>\n\nGraesser, A. C. (in press). Conversations with AutoTutor help students learn. International Journal of Artificial Intelligence in Education<\/i>. <\/span>\n\nGraesser, A. C., Lu, S., Jackson, G. T., Mitchell, H. H., Ventura, M., Olney, A., & Louwerse, M. M. (2004). AutoTutor: A tutor with dialogue in natural language. Behavior Research Methods, Instruments, & Computers, 36<\/i>(2), 180\u2013192. <\/span>\n\nGraesser, A. C., McNamara, D. S., & Kulikowich, J. M. (2011). Coh Metrix: Providing multilevel analyses of text characteristics. Educational Researcher, 40<\/i>, 223\u2013234. <\/span>\n\nGraesser, A. C., McNamara, D. S., & VanLehn, K. (2005). Scaffolding deep comprehension strategies through Point & Query, AutoTutor, and iSTART. Educational Psychologist, 40<\/i>, 225\u2013234. <\/span>\n\nGraesser, A. C., & Person, N. K. (1994). Question asking during tutoring. American Educational Research Journal, 31<\/i>(1), 104\u2013137. <\/span>\n\nJackson, G. T., Allen, L. K., & McNamara, D. S. (2016). Common Core TERA: Text Ease and Readability Assessor. In S. A. Crossley & D. S. McNamara (Eds.), Adaptive educational technologies for literacy instruction<\/i>. New York: Taylor & Francis, Routledge. <\/span>\n\nJackson, G. T., Guess, R. H., & McNamara, D. S. (2010). Assessing cognitively complex strategy use in an untrained domain. Topics in Cognitive Science, 2<\/i>, 127\u2013137. <\/span>\n\nJarvis, S., Bestgen, Y., Crossley, S. A., Granger, S., Paquot, M., Thewissen, J., & McNamara, D. S. (2012). The comparative and combined contributions of n-grams, Coh-Metrix indices and error types in the L1 classification of learner texts. In S. Jarvis & S. A. Crossley (Eds.), Approaching language transfer through text classification: Explorations in the detection-based approach <\/i>(pp. 154\u2013177). Bristol, UK: Multilingual Matters. <\/span>\n\nJohnson-Glenberg, M. C. (2007). Web-based reading comprehension instruction: Three studies of 3D-readers. In D. McNamara (Ed.), Reading comprehension strategies: Theory, interventions, and technologies <\/i>(pp. 293\u2013324). Mahwah, NJ: Lawrence Erlbaum Publishers. <\/span>\n\nJurafsky, D., & Martin, J. H. (2000). Speech and language processing: An introduction to natural language processing, computational linguistics and speech recognition. Upper Saddle River, NJ: Prentice Hall. <\/span>\n\nJurafsky, D., & Martin, J. H. (2008). Speech and language processing: An introduction to natural language processing, computational linguistics and speech recognition, 2nd ed. Upper Saddle River, NJ: Prentice Hall. <\/span>\n\nKoller, D., Ng, A., Do, C., & Chen, Z. (2013). Retention and intention in massive open online courses: In depth. Educause Review, 48<\/i>(3), 62\u201363. <\/span>\n\nKoschmann, T. (1999). Toward a dialogic theory of learning: Bakhtin's contribution to understanding learning in settings of collaboration. In C. M. Hoadley & J. Roschelle (Eds.), Proceedings of the 1999 Conference on Computer Support for Collaborative Learning <\/i>(CSCL\u201999), 12\u201315 December 1999, Palo Alto, California (pp. 308\u2013313). International Society of the Learning Sciences. <\/span>\n\nKyle, K., & Crossley, S. A. (2015). Automatically assessing lexical sophistication: Indices, tools, findings, and application. TESOL Quarterly, 49<\/i>(4), 757\u2013786. <\/span>\n\nLandauer, T. K., & Dumais, S. T. (1997). A solution to Plato's problem: The latent semantic analysis theory of acquisition, induction, and representation of knowledge. Psychological Review, 104<\/i>(2), 211\u2013240. <\/span>\n\nLandauer, T. K., Laham, D., & Foltz, P. W. (2003). Automated scoring and annotation of essays with the Intelligent Essay Assessor. In M. D. Shermis & J. Burstein (Eds.), Automated essay scoring: A cross-disciplinary perspective <\/i>(pp. 87\u2013112). Mahwah, NJ: Lawrence Erlbaum.<\/span>\n\nLandauer, T., McNamara, D. S., Dennis, S., & Kintsch, W. (Eds.). (2007). Handbook of latent semantic analysis<\/i>. Mahwah, NJ: Lawrence Erlbaum. <\/span>\n\nLandauer, T. K., Kireyev, K., & Panaccione, C. (2011). Word maturity: A new metric for word knowledge. Scientific Studies of Reading, 15<\/i>(1), 92\u2013108. <\/span>\n\nLouwerse, M. M., McCarthy, P. M., McNamara, D. S., & Graesser, A. C. (2004). Variation in language and cohesion across written and spoken registers. In K. Forbus, D. Gentner, & T. Regier (Eds.), Proceedings of the 26th Annual Conference of the Cognitive Science Society <\/i>(CogSci 2004), 4\u20137 August 2004, Chicago, IL, USA (pp. 843\u2013848). Mahwah, NJ: Lawrence Erlbaum. <\/span>\n\nMcCarthy, P. M., Briner, S. W., Rus, V., & McNamara, D. S. (2007). Textual signatures: Identifying text-types using latent semantic analysis to measure the cohesion of text structures. In A. Kao & S. Poteet (Eds.), Natural language processing and text mining <\/i>(pp. 107\u2013122). London: Springer-Verlag UK. <\/span>\n\nMcKeown, M. G., Beck, I. L., & Blake, R. G. K. (2009). Rethinking reading comprehension instruction: A comparison of instruction for strategies and content approaches. Reading Research Quarterly, 44<\/i>, 218\u2013253. <\/span>\n\nMcNamara, D. S. (2011). Computational methods to extract meaning from text and advance theories of human cognition. Topics in Cognitive Science, 2<\/i>, 1\u201315. <\/span>\n\nMcNamara, D. S., Boonthum, C., Levinstein, I. B., & Millis, K. (2007). Evaluating self-explanations in iSTART: Comparing word-based and LSA algorithms. In T. Landauer, D. S. McNamara, S. Dennis, & W. Kintsch (Eds.), Handbook of latent semantic analysis <\/i>(pp. 227\u2013241). Mahwah, NJ: Lawrence Erlbaum. <\/span>\n\nMcNamara, D. S., Crossley, S. A., & Roscoe, R. D. (2013). Natural language processing in an intelligent writing strategy tutoring system. Behavior Research Methods, 45<\/i>, 499\u2013515. <\/span>\n\nMcNamara, D. S., Graesser, A. C., McCarthy, P., & Cai, Z. (2014). Automated evaluation of text and discourse with Coh-Metrix<\/i>. Cambridge, UK: Cambridge University Press. <\/span>\n\nMcNamara, D. S., & Kintsch, W. (1996). Learning from text: Effects of prior knowledge and text coherence. Discourse Processes, 22<\/i>, 247\u2013288. <\/span>\n\nMcNamara, D. S., Levinstein, I. B., & Boonthum, C. (2004). iSTART: Interactive strategy trainer for active reading and thinking. Behavioral Research Methods, Instruments, & Computers, 36<\/i>, 222\u2013233. <\/span>\n\nMcNamara, D. S., Ozuru, Y., Graesser, A. C., & Louwerse, M. (2006). Validating Coh-Metrix. In R. Sun & N. Miyake (Eds.), Proceedings of the 28th Annual Conference of the Cognitive Science Society <\/i>(CogSci 2006), 26\u201329 July 2006, Vancouver, British Columbia, Canada (pp. 573\u2013578). Austin, TX: Cognitive Science Society. <\/span>\n\nMcNamara, D. S., Raine, R., Roscoe, R., Crossley, S. A, Jackson, G. T., Dai, J., Cai, Z., Renner, A., Brandon, R., Weston, J., Dempsey, K., Carney, D., Sullivan, S., Kim, L., Rus, V., Floyd, R., McCarthy, P. M., & Graesser, A. C. (2012). The Writing-Pal: Natural language algorithms to support intelligent tutoring on writing strategies. In P. M. McCarthy & C. Boonthum-Denecke (Eds.), Applied natural language processing and content analysis: Identification, investigation, and resolution <\/i>(pp. 298\u2013311). Hershey, PA: IGI Global. <\/span>\n\nMikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word representation in vector space. In Workshop at ICLR<\/i>. Scottsdale, AZ. https:\/\/arxiv.org\/abs\/1301.3781 <\/span>\n\nMiller, G. A., Beckwith, R., Fellbaum, C., Gross, D., & Miller, K. J. (1990). Introduction to WordNet: An on-line lexical database. International Journal of Lexicography, 3<\/i>(4), 235\u2013244. <\/span>\n\nMoon, S., Potdar, S., & Martin, L. (2014). Identifying student leaders from MOOC discussion forums through language influence. EMNLP 2014<\/i>, 15. http:\/\/www.aclweb.org\/anthology\/W14-4103 <\/span>\n\nPennebaker, J. W., Booth, R. J., & Francis, M. E. (2007). Linguistic inquiry and word count: LIWC [Computer software]. Austin, TX: liwc.net. <\/span>\n\nPennebaker, J. W., Boyd, R. L., Jordan, K., & Blackburn, K. (2015). The development and psychometric properties of LIWC2015. UT Faculty\/Researcher Works. https:\/\/repositories.lib.utexas.edu\/bitstream\/handle\/2152\/31333\/LIWC2015_LanguageManual.pdf?sequence=3<\/span>\n\nRoscoe, R. D., & McNamara, D. S. (2013). Writing Pal: Feasibility of an intelligent writing strategy tutor in the high school classroom. Journal of Educational Psychology, 105<\/i>, 1010\u20131025. <\/span>\n\nRudner, L. M., Garcia, V., & Welch, C. (2006). An evaluation of IntelliMetric' essay scoring system. The Journal of Technology, Learning and Assessment, 4<\/i>(4). https:\/\/ejournals.bc.edu\/ojs\/index.php\/jtla\/article\/download\/1651\/1493 <\/span>\n\nSeaton, D. T., Bergner, Y., Chuang, I., Mitros, P., & Pritchard, D. E. (2014). Who does what in a massive open online course? Communications of the ACM, 57<\/i>(4), 58\u201365. <\/span>\n\nShermis, M. D., Burstein, J., Higgins, D., & Zechner, K. (2010). Automated essay scoring: Writing assessment and instruction. International Encyclopedia of Education, 4<\/i>, 20\u201326. <\/span>\n\nStahl, G. (2006). Group cognition: Computer support for building collaborative knowledge (pp. 451\u2013473). Cambridge, MA: MIT Press. <\/span>\n\nTeplovs, C. (2008). The knowledge space visualizer: A tool for visualizing online discourse. In G. Kanselaar, V. Jonker, P. A. Kirschner, & F. Prins (Eds.), Proceedings of the International Society of the Learning Sciences 2008: Cre8 a learning world. <\/i>Utrecht, NL: International Society of the Learning Sciences. http:\/\/chris.ikit.org\/ksv2.pdf <\/span>\n\nTrausan-Matu, S., Rebedea, T., Dragan, A., & Alexandru, C. (2007). Visualisation of learners' contributions in chat conversations. In J. Fong & F. L. Wang (Eds.), Blended learning <\/i>(pp. 217\u2013226). Singapur: Pearson\/Prentice Hall. <\/span>\n\nTrausan-Matu, S., Stahl, G., & Sarmiento, J. (2007). Supporting polyphonic collaborative learning. Indiana University Press, E-service Journal, 6<\/i>(1), 58\u201374. <\/span>\n\nTuring, A. M. (1950). Computing machinery and intelligence. Mind, 59, 433\u2013460. http:\/\/www.loebner.net\/ Prizef\/TuringArticle.html <\/span>\n\nValenti, S., Neri, F., & Cucchiarelli, A. (2003). An overview of current research on automated essay grading. Journal of Information Technology Education: Research, 2<\/i>(1), 319\u2013330. <\/span>\n\nVarner, L. K., Roscoe, R. D., & McNamara, D. S. (2013). Evaluative misalignment of 10th-grade student and teacher criteria for essay quality: An automated textual analysis. Journal of Writing Research, 5<\/i>, 35\u201359. <\/span>\n\nWeigle, S.C. (2013). English as a second language writing and automated essay evaluation. In M. D. Shermis & J. Burstein (Eds.), Handbook of automated essay evaluation: Current applications and new directions <\/i>(pp. 36\u201354). London: Routledge. <\/span>\n\nWen, M., Yang, D., & Ros\u00e9, C. P. (2014a). Linguistic reflections of student engagement in massive open online courses. In E. Adar & P. Resnick (Eds.), Proceedings of the 8th International AAAI Conference on Weblogs and Social Media <\/i>(ICWSM\u201914), 1\u20134 June 2014, Ann Arbor, Michigan, USA. Palo Alto, CA: AAAI Press. http:\/\/www. aaai.org\/ocs\/index.php\/ICWSM\/ICWSM14\/paper\/viewFile\/8057\/8153 <\/span>\n\nWen, M., Yang, D., & Ros\u00e9, C. P. (2014b). Sentiment analysis in MOOC discussion forums: What does it tell us? In J. Stamper, Z. Pardos, M. Mavrikis, & B. M. McLaren (Eds.), Proceedings of the 7th International Conference on Educational Data Mining <\/i>(EDM2014), 4\u20137 July, London, UK (pp. 185\u2013192). International Educational Data Mining Society. <\/span>\n\nWilson, M.D. (1988). The MRC psycholinguistic database: Machine-readable dictionary (Version 2). Behavioral Research Methods, Instruments, and Computers, 201<\/i>, 6\u201311. <\/span>\n\nWong, B. Y. L. (1985). Self-questioning instructional research: A review. Review of Educational Research, 55<\/i>, 227\u2013268. <\/span>\n

Xi, X. (2010). Automated scoring and feedback systems: Where are we and where are we heading? Language Testing, 27<\/i>(3), 291\u2013300.<\/p>\n\n\n

\n\n
\n\n1<\/a> orj. descriptive statistics<\/span>\n\n<\/div>\n
\n\n2<\/a> orj. discriminant<\/span>\n\n<\/div>\n","rendered":"

Danielle S. McNamara1<\/sup>, Laura K. Allen1<\/sup>, Scott A. Crossley2<\/sup>, Mihai Dascalu3<\/sup>, Cecile A. Perret4 <\/sup><\/span><\/span><\/p>\n

1<\/sup>Psikoloji B\u00f6l\u00fcm\u00fc, Arizona Devlet \u00dcniversitesi, ABD <\/span><\/span><\/p>\n

2<\/sup> Uygulamal\u0131 Dilbilim ve ESL B\u00f6l\u00fcm\u00fc, Georgia Devlet \u00dcniversitesi, ABD<\/span><\/span><\/p>\n

3 <\/sup>Bilgisayar Bilimleri B\u00f6l\u00fcm\u00fc, B\u00fckre\u015f Politeknik \u00dcniversitesi, Romanya <\/span><\/span><\/p>\n

4<\/sup> \u00d6\u011fretme ve \u00d6\u011frenme Bilim Enstit\u00fcs\u00fc, Arizona Devlet \u00dcniversitesi, ABD <\/span><\/span><\/p>\n

DOI: 10.18608\/hla17.008<\/span><\/span><\/p>\n

\u00d6Z<\/h2>\n

Dil, bilgiyi anlamak ve ili\u015fki kurmak i\u00e7in bir kanal oldu\u011fundan e\u011fitim alan\u0131 i\u00e7in merkezi bir \u00f6neme sahiptir. Bu nedenle, \u00f6\u011frenme analiti\u011fi alan\u0131ndaki ara\u015ft\u0131rmac\u0131lar, dili do\u011fru ve etkili bir \u015fekilde analiz etmek i\u00e7in geli\u015ftirilen y\u00f6ntemlerden yararlanabilir. Do\u011fal dil i\u015fleme (DD\u0130) teknikleri b\u00f6yle bir yol sa\u011flayabilir. DD\u0130 teknikleri, belirli g\u00f6revlerle ilgili olarak dilin farkl\u0131 y\u00f6nlerinin bilgi i\u015flemsel analizlerini sa\u011flamak i\u00e7in kullan\u0131l\u0131r. Bu b\u00f6l\u00fcmde, yazarlar s\u00f6ylemin anla\u015f\u0131lmas\u0131 i\u00e7in kullan\u0131labilecek bir\u00e7ok DD\u0130 ara\u00e7lar\u0131 ve bu e\u011fitim ara\u00e7lar\u0131n\u0131n baz\u0131 uygulamalar\u0131n\u0131 tart\u0131\u015fmaktad\u0131r. Bu ara\u00e7lar\u0131n ana oda\u011f\u0131, insanlar ve bilgisayarlar aras\u0131ndaki etkile\u015fimi veya insan-bilgisayar etkile\u015fimini sa\u011flamak i\u00e7in insan dili girdisinin otomatik olarak yorumlanmas\u0131d\u0131r. Bu nedenle, ara\u00e7lar metinleri anlamak i\u00e7in \u00f6nemli olan tutarl\u0131l\u0131k, s\u00f6zdizimsel karma\u015f\u0131kl\u0131k, kelime \u00e7e\u015fitlili\u011fi ve anlamsal benzerlik gibi \u00e7e\u015fitli dil \u00f6zelliklerini \u00f6l\u00e7er. Yazarlar, b\u00f6l\u00fcm\u00fc DD\u0130 ara\u00e7lar\u0131n\u0131 (yani, Ak\u0131ll\u0131 \u00d6\u011fretim Sistemleri, KA\u00c7D’ler ve Bilgisayar Destekli \u0130\u015fbirlikli \u00d6\u011frenme) kullanan bilgisayar tabanl\u0131 \u00f6\u011frenme ortamlar\u0131 ve bu t\u00fcr ara\u00e7lar\u0131n gelecekteki ara\u015ft\u0131rmalarda nas\u0131l kullan\u0131labilece\u011fi tart\u0131\u015fmas\u0131yla sonland\u0131rmaktad\u0131r.<\/span><\/p>\n

Anahtar Kelimeler<\/b>:Do\u011fal dil i\u015fleme (DD\u0130), dil, bilgi i\u015flemsel dil bilimi, dil \u00f6zellikleri, otomatik yazma de\u011ferlendirmesi, ak\u0131ll\u0131 \u00f6\u011fretici sistemleri, BD\u0130\u00d6, KA\u00c7D<\/span><\/p>\n

Dil d\u00fc\u015f\u00fcncelerimizi d\u0131\u015far\u0131ya yans\u0131tman\u0131n bir arac\u0131d\u0131r. Kendimizi ba\u015fkalar\u0131na ifade etmemize, d\u00fcnyay\u0131 manip\u00fcle etmemizi ve \u00e7evrede bulunan nesneleri etiketlememizi sa\u011flar. Dil, d\u00fc\u015f\u00fcncelerimizi i\u00e7sel olarak in\u015fa etmemizi ve yeniden yap\u0131land\u0131rmam\u0131z\u0131 sa\u011flar; d\u00fc\u015f\u00fcncelerimizi temsil edebilir ve onlar\u0131 d\u00f6n\u00fc\u015ft\u00fcrmemize imkan tan\u0131r. Sosyal deneyimler kurmam\u0131z\u0131 ve \u015fekillendirmemizi sa\u011flar. Dil, d\u00fcnyay\u0131 anlamak ve etkile\u015fimde bulunmak i\u00e7in bir kanal sa\u011flar.<\/p>\n

Dil ya\u015fam\u0131m\u0131zda, d\u00fc\u015f\u00fcncelerimizde, ileti\u015fimimizde, okuduklar\u0131m\u0131zda ve yazd\u0131klar\u0131m\u0131zda ve di\u011ferleriyle etkile\u015fimlerimizde her yerdedir. Dil, e\u011fitim i\u00e7in ayn\u0131 \u015fekilde merkezidir. \u00d6\u011fretenler olarak hedefimiz, \u00f6\u011frencilere yeni bilgileri \u00f6\u011frenme, bunlar\u0131 \u00f6z\u00fcmseme ve b\u00fct\u00fcnle\u015ftirme f\u0131rsat\u0131na sahip olacak \u015fekilde bilgileri iletmektir. \u00d6\u011frenciler bilgiyi iletmek i\u00e7in kullan\u0131lan dili anlama ve daha sonra bu bilgileri bireyler olarak, gruplar halinde, birbirleriyle ve \u00f6\u011fretenlerle e\u015fg\u00fcd\u00fcm halinde in\u015fa etmeyi bildikleri \u015feylerle ili\u015fkilendirmekle g\u00f6revlidirler.<\/p>\n

Dil, ya\u015fam\u0131m\u0131zda ve e\u011fitimde \u00f6nemli roller oynar ve bu nedenle bu rolleri ve \u00e7\u0131kt\u0131lar\u0131 tan\u0131mak ve anlamak \u00f6nemlidir. Metin ve s\u00f6ylem analizi, dil kullan\u0131m\u0131yla ilgili karma\u015f\u0131k s\u00fcre\u00e7leri anlamak i\u00e7in bir yol sa\u011flar. S\u00f6ylem analistleri, yaz\u0131l\u0131 metin ve s\u00f6zl\u00fc s\u00f6ylem i\u00e7erisindeki yap\u0131lar\u0131, \u00f6r\u00fcnt\u00fcleri ve onlar\u0131n davran\u0131\u015flarla, psikolojik s\u00fcre\u00e7lerle, bili\u015fsel ve sosyal etkile\u015fimlerle ili\u015fkilerini sistematik olarak inceler. Nitekim, metin ve s\u00f6ylem analizi dil hakk\u0131nda bir bilgi hazinesi sa\u011flam\u0131\u015ft\u0131r.<\/p>\n

Geleneksel olarak, s\u00f6ylem analizi zahmetlidir. \u0130lk olarak, \u00f6rne\u011fin, anlaml\u0131 dil birimleri tan\u0131mlan\u0131r ve b\u00f6l\u00fcmlere ayr\u0131l\u0131r (\u00f6r. t\u00fcmceler, ifadeler) ve sonra uzmanlar bu birimleri kodlar (yani, belirli bir analiz i\u00e7in). Ard\u0131ndan, bu dil birimlerinin do\u011fas\u0131 ve \u00e7\u0131kt\u0131lar\u0131 aras\u0131ndaki potansiyel ili\u015fkiler de\u011ferlendirilir. Bireyler aras\u0131nda binlerce ifade ve al\u0131\u015fveri\u015fin oldu\u011fu b\u00fcy\u00fck veri d\u00fcnyas\u0131nda, dili elle-kodlama neredeyse imk\u00e2ns\u0131zd\u0131r. B\u00fcy\u00fck veri derlemleri, dili daha geni\u015f ve daha anlaml\u0131 bir \u00f6l\u00e7ekte anlamak i\u00e7in kap\u0131lar\u0131 a\u00e7ar ancak s\u00f6ylem analizine geleneksel yakla\u015f\u0131mlar uygulanabilir ve elveri\u015fli de\u011fildir. Do\u011fal dil i\u015flemeden<\/i><\/span> (DD\u0130) bir \u00e7\u00f6z\u00fcm elde edilebilir.<\/p>\n

DD\u0130, insan dilinin bilgisayar dili kullanarak analizidir ve s\u00f6ylem \u00e7\u00f6z\u00fcmlemesini otomatikle\u015ftirmenin yolunu g\u00f6sterir. DD\u0130 terimi, bilgisayar dillerinin kullan\u0131m\u0131 ve analizinin aksine, do\u011fal insan dilinin analizi oldu\u011fu i\u00e7in kullan\u0131lm\u0131\u015ft\u0131r. Do\u011fal dili i\u015flemek i\u00e7in \u00e7e\u015fitli otomatik ara\u00e7lar kullan\u0131labilir. Nitekim, DD\u0130 ara\u00e7lar\u0131n\u0131n say\u0131s\u0131 ve g\u00fcc\u00fc 1990’lar\u0131n ortas\u0131ndan bu yana d\u00fczenli olarak artm\u0131\u015ft\u0131r (Jurafsky ve Martin, 2000, 2008). Dolay\u0131s\u0131yla \u00f6\u011frenme analiti\u011fi ve veri madencili\u011fi alan\u0131ndaki etkileri ve kullan\u0131mlar\u0131, katlamal\u0131 olarak olmasa da giderek artmaktad\u0131r. Bu b\u00f6l\u00fcm, ara\u015ft\u0131rmac\u0131lar ve e\u011fitimciler i\u00e7in dili bilgi i\u015flemsel olarak analiz etmek i\u00e7in mevcut olan ve \u00f6zellikle de e\u011fitim alan\u0131ndaki kullan\u0131mlar\u0131na odaklanan \u00e7e\u015fitli ara\u00e7lar\u0131 a\u00e7\u0131klamaktad\u0131r.<\/p>\n

DO\u011eAL D\u0130L \u0130\u015eLEME<\/h2>\n

Bilgii\u015flemsel dilbilim modelleri geli\u015fimine odaklanan bir disiplindir. DD\u0130 ara\u00e7 ve teknikleri genellikle bilgi i\u015flemsel dilbilimi alan\u0131nda geli\u015ftirilen teoriler, modeller ve algoritmalar taraf\u0131ndan y\u00f6nlendirilir ancak DD\u0130 ara\u00e7lar\u0131n\u0131n temel amac\u0131 insan dili girdilerinin otomatik olarak yorumlanmas\u0131d\u0131r. B\u00f6yle bir \u00e7aba, dilbilim, bilgisayar bilimi, psikoloji ve e\u011fitim gibi disiplinleri birle\u015ftiren disiplinleraras\u0131 bak\u0131\u015f a\u00e7\u0131lar\u0131 \u00fczerinde durur. DD\u0130’nin Turing’e (1950) dayanan bir ge\u00e7mi\u015fi olsa da mevcut DD\u0130 algoritmalar\u0131n\u0131n \u00e7o\u011fu, DD\u0130 ara\u00e7lar\u0131 ve veri madencili\u011finin birle\u015fimi kullan\u0131larak geli\u015ftirilmi\u015ftir. Bilgisayar veya veri bilimcileri taraf\u0131ndan s\u0131kl\u0131kla kullan\u0131lan DD\u0130 yaz\u0131l\u0131m\u0131 ile bu b\u00f6l\u00fcmde sunulan ara\u00e7lar aras\u0131nda en ba\u015ftan net bir ayr\u0131m yap\u0131lmal\u0131d\u0131r. DD\u0130’ye ili\u015fkin ara\u015ft\u0131rmalar\u0131n b\u00fcy\u00fck bir \u00e7o\u011funlu\u011fu y\u00fczeysel seviyede metin i\u015flemeye odaklanm\u0131\u015ft\u0131r (\u00f6r. makine \u00e7evirisi) ve mevcut ara\u00e7lar sonu\u00e7 olarak do\u011fru kelime -ve c\u00fcmle- d\u00fczeyi metin i\u015flemenin temel rol\u00fcn\u00fc vurgulamaktad\u0131r. Bu b\u00f6l\u00fcmdeki amac\u0131m\u0131z, \u00f6\u011frenme analiti\u011fi ba\u011flam\u0131nda DD\u0130’ye odaklanmakt\u0131r. Bu nedenle, bu y\u00fczey d\u00fczeyinde g\u00f6revlerin \u00f6tesine ge\u00e7en, kelime dizinlerini hesaplamak i\u00e7in geli\u015ftirilen ve e\u011fitim ba\u011flam\u0131nda daha \u00f6nemli olabilecek bilgileri sunan ara\u00e7lara odaklan\u0131yoruz. Bilhassa, birden fazla metin d\u00fczeyi hakk\u0131nda bilgi sa\u011flayan bir DD\u0130 teknikleri alt k\u00fcmesini betimliyoruz. Bu ara\u00e7lar s\u00f6ylemdeki s\u00f6zc\u00fcklerden ba\u015flar, belirli kelime \u00f6zelliklerini \u00e7\u0131kar\u0131r ve ard\u0131ndan anlam yap\u0131s\u0131n\u0131 ve s\u00f6ylem yap\u0131s\u0131n\u0131 dikkate alarak veri s\u00f6zl\u00fc\u011f\u00fcn\u00fcn \u00f6tesine ge\u00e7er. Amac\u0131m\u0131z, mevcut t\u00fcm y\u00f6ntemlere genel bir bak\u0131\u015ftan ziyade, birka\u00e7 ortak tekni\u011fe dair \u00f6rnekler sunmakt\u0131r. Bu y\u00f6ntemleri,do\u011frudan analiz birimleri olarak kelimelere odaklananlar ve kelimelerin \u00f6zelliklerine odaklananlar olarak grupland\u0131r\u0131yoruz.<\/p>\n

Kelimeler<\/h3>\n

DD\u0130’ye y\u00f6nelik bir yakla\u015f\u0131m, dilde kullan\u0131lan kelimeleri do\u011frudan analiz etmektir. \u00d6rne\u011fin, bir metindeki belirli s\u00f6zc\u00fck t\u00fcrlerinin g\u00f6r\u00fclme s\u0131kl\u0131\u011f\u0131n\u0131 hesaplamak, \u00e7e\u015fitli ba\u011flamlarda kullan\u0131lan dilin do\u011fas\u0131 ve amac\u0131na dair iyi bir y\u00f6ntem olu\u015fturabilir. Bu genellikle “s\u00f6zc\u00fck \u00e7antas\u0131” yakla\u015f\u0131m\u0131 olarak adland\u0131r\u0131l\u0131r. Bu yakla\u015f\u0131m\u0131 kullanan ara\u00e7lardan biri, Pennebaker ve meslekta\u015flar\u0131, taraf\u0131ndan geli\u015ftirilen Dilbilimsel Sorgu ve Kelime Say\u0131s\u0131 (DSKS) sistemidir (Pennebaker, Booth ve Francis, 2007; Pennebaker, Boyd, Jordan ve Blackburn, 2015; bk. http: \/\/liwc.wpengine.com). DSKS’nin 2007 s\u00fcr\u00fcm\u00fc kabaca 80 kelime kategorisi sa\u011flar, fakat ayn\u0131 zamanda bu kelime kategorilerini daha geni\u015f boyutlarda grupland\u0131r\u0131r. Daha geni\u015f boyutlar\u0131n \u00f6rnekleri dil bi\u00e7imleri (\u00f6r. zamirler, ge\u00e7mi\u015f zamandaki kelimeler, ters ifadeler), sosyal s\u00fcre\u00e7ler, duygusal s\u00fcre\u00e7ler ve bili\u015fsel s\u00fcre\u00e7lerdir. \u00d6rne\u011fin, bili\u015fsel s\u00fcre\u00e7ler, i\u00e7g\u00f6r\u00fc (\u00f6r. d\u00fc\u015f\u00fcn, bil, g\u00f6z \u00f6n\u00fcne al), nedensellik (\u00f6r. \u00e7\u00fcnk\u00fc, etki, dolay\u0131s\u0131yla) ve kesinlik (\u00f6r. her zaman, asla) gibi alt kategorileri i\u00e7erir. DSKS, her bir kelime kategorisine ait kelimelerin say\u0131s\u0131n\u0131 sayar ve kategorideki kelimelerin say\u0131s\u0131n\u0131 metindeki kelimelerin toplam say\u0131s\u0131na b\u00f6lerek bir oran sunar.<\/p>\n

Benzer bir yakla\u015f\u0131m, karakterlerin veya kelimelerin gruplar\u0131 gibi n-gramlar\u0131n\u0131 tan\u0131mlamakt\u0131r; burada n, gruba d\u00e2hil edilen gramlar\u0131n say\u0131s\u0131n\u0131 belirtir (\u00f6r. \u0130ki gramlar, iki kelimeli gruplara at\u0131fta bulunur). N-gram analizleri, metinlerdeki kelime dizilimlerin olas\u0131l\u0131k da\u011f\u0131l\u0131mlar\u0131n\u0131 hesaplar ve bir metin grubuna ortak olan veya belirli bir metin veya metin dizileri i\u00e7in farkl\u0131 olan kelimeler hakk\u0131nda bilgi sa\u011flayabilir (\u00f6r. Jarvis vd., 2012). N-gram analizlerinin avantajlar\u0131, basitlikleri ve bir metnin spesifik i\u00e7eri\u011fi, bir metnin dil ve s\u00f6zdizimsel \u00f6zellikleri ve bu \u00f6zellikler aras\u0131ndaki ili\u015fkiler hakk\u0131nda bilgi sa\u011flama potansiyelini i\u00e7erir (Crossley ve Louwerse, 2007).<\/p>\n

Kelimelerin \u00d6zellikleri<\/h3>\n

Kelimelerin olu\u015fumunu ve kelime gruplar\u0131n\u0131 hesaplamak metnin a\u00e7\u0131k i\u00e7eri\u011fini g\u00f6z \u00f6n\u00fcne al\u0131r. Alternatif bir yakla\u015f\u0131m, bir metindeki kelimelerin ve c\u00fcmlelerin \u00f6zelliklerinin hesaplanmas\u0131n\u0131 i\u00e7erir. Bu t\u00fcr bir teknik, kelimelerin arkas\u0131ndaki gizli anlam\u0131 elde etmek i\u00e7indir (McNamara, 2011). Bunu yapmak i\u00e7in \u00e7ok say\u0131da algoritma olsa da en iyi bilinen ve belki de ilk olan\u0131 \u00f6rt\u00fck semantik analiz (\u00d6SA; Landauer ve Dumais, 1997; Landauer, McNamara, Dennis ve Kintsch, 2007; bk. lsa.colorado.edu) dir. \u00d6SA 1990’lar\u0131n ortalar\u0131nda ortaya \u00e7\u0131kt\u0131, b\u00fcy\u00fck metin g\u00f6vdelerinden semantik anlam \u00e7\u0131karmak ve b\u00fcy\u00fck ve k\u00fc\u00e7\u00fck metin \u00f6rneklerini semantik benzerliklerle kar\u015f\u0131la\u015ft\u0131rmak i\u00e7in bir ara\u00e7 sundu. Bu yakla\u015f\u0131m DD\u0130’de bir devrim yaratmak i\u00e7in benzersiz bir potansiyel sa\u011flad\u0131. \u00d6SA, geni\u015f bir belge k\u00fcmesinde kelimelerin var olu\u015funu temsil eden bir matrisi s\u0131k\u0131\u015ft\u0131rmak (yani \u00e7arpanlara ay\u0131rmak) i\u00e7in tekil de\u011fer ayr\u0131\u015ft\u0131rmas\u0131 kullanan matematiksel, istatistiksel bir tekniktir. \u00d6SA’y\u0131 y\u00f6nlendiren temel varsay\u0131m kelimelerin anlamlar\u0131n\u0131n onlarla birlikte olanlar taraf\u0131ndan yakaland\u0131\u011f\u0131 idi. \u00d6rne\u011fin, “veri” kelimesi, “hesaplamalar”, “madencilik”, “bilgisayar” ve “matematik” gibi ayn\u0131 i\u015flevsel ba\u011flamdaki kelimelerle b\u00fcy\u00fck \u00f6l\u00e7\u00fcde ili\u015fkilendirilecektir. Bu kelimeler veri ile ayn\u0131 anlama gelmemektedir. Daha do\u011frusu, bu kelimeler genellikle benzer ba\u011flamlarda olu\u015ftu\u011fu i\u00e7in verilerle ilgilidir. \u00d6SA kelimeler, c\u00fcmleler ve paragraflar aras\u0131ndaki semantik benzerliklerin hesaplanmas\u0131n\u0131 sa\u011flayarak, metinde anlam simulasyonunun kap\u0131lar\u0131n\u0131 a\u00e7t\u0131 (McNamara, 2011). \u00d6SA, basit kelime \u00f6rt\u00fc\u015fme \u00f6nlemlerinin yeterli olmad\u0131\u011f\u0131 bir problem olan anlaml\u0131l\u0131k sorununu (\u00f6r. bir metnin ba\u015fka bir metinle veya \u00e7ekirdek bir kavramla ne derece alakal\u0131 oldu\u011fu) ba\u015far\u0131yla ele alan ilk kelime temelli yakla\u015f\u0131m olarak d\u00fc\u015f\u00fcn\u00fclebilir. \u00d6SA’n\u0131n \u00f6tesine ge\u00e7en \u00e7ok say\u0131da yakla\u015f\u0131m olsa da (genel bir bak\u0131\u015f i\u00e7in bk. McNamara, 2011), \u00d6SA, kelime anlam\u0131n\u0131 modellemek ve anlam bilim ve metin uyumu bak\u0131m\u0131ndan i\u00e7g\u00f6r\u00fc sa\u011flamak i\u00e7in bir\u00e7ok ba\u011flamda kullan\u0131lan ortak bir yakla\u015f\u0131m olmaya devam etmektedir (\u00f6r. Landauer vd., 2007; McNamara, Graesser, McCarthy ve Cai, 2014).<\/p>\n

Dilin bariz bir \u00f6zelli\u011fi anlam\u0131d\u0131r ancak konu\u015fma b\u00f6l\u00fcmleri (\u00f6r. fiil, isim), s\u00f6zdizimi, psikolojik y\u00f6nler (\u00f6r. somutluk, anlaml\u0131l\u0131k) ve aras\u0131ndaki ili\u015fkiler, metindeki fikirler (\u00f6r. uyum) gibi bir \u00e7ok di\u011fer \u00f6zellikler dilbilimsel analizlerden t\u00fcretilebilir. Coh-Metrix, ilk olarak 2003 y\u0131l\u0131nda piyasaya s\u00fcr\u00fclen, metnin dil, psikolojik ve semantik \u00f6zelliklerini \u00e7\u0131karmak i\u00e7in dil hakk\u0131nda bir\u00e7ok bilgi kayna\u011f\u0131 kullanan otomatik bir dil analiz arac\u0131 \u00f6rne\u011fidir (McNamara vd., 2014; cohmetrix.com). Coh-Metrix, \u0130ngilizce dili ile ilgili bilgileri \u00d6SA, MRC Psikodilbilimsel Veri Taban\u0131, WordNet ve CELEX gibi kelime s\u0131kl\u0131\u011f\u0131 dizinleri ile s\u00f6zdizimsel ayr\u0131\u015ft\u0131r\u0131c\u0131lar gibi \u00e7e\u015fitli kaynaklardan uyarlar ve birle\u015ftirir. \u00d6rne\u011fin, MRC Psikodilbilimsel Veri Taban\u0131, kelimeler hakk\u0131nda psikodilbilimsel bilgi sa\u011flar (Wilson, 1988) ve WordNet, kelimelerin dil ve anlam \u00f6zellikleri ve ayr\u0131ca kelimeler (Miller, Beckwith, Fellbaum, Gross ve Miller, 1990) aras\u0131ndaki anlamsal ili\u015fkileri de sa\u011flar. Coh-Metrix ayr\u0131ca, yaz\u0131l\u0131 veya s\u00f6zl\u00fc metinlerin \u00e7ok boyutlu bir analiz \u00fcretmek i\u00e7in s\u00f6zc\u00fck s\u0131kl\u0131\u011f\u0131 ve c\u00fcmle uzunlu\u011fu gibi metin kalitesinin basit \u00f6zellikleri, tutarl\u0131l\u0131k ve s\u00f6zdizimsel karma\u015f\u0131kl\u0131k gibi daha karma\u015f\u0131k \u00f6zellikler sayesinde dilin \u00e7e\u015fitli y\u00f6nleriyle ilgili dil indekslerini de hesaplar. (McNamara, Ozuru, Graesser ve Louwerse, 2006). Coh \u2013 Metrix, tan\u0131mlay\u0131c\u0131 dizinlerle (\u00f6r. kelimelerin uzunlu\u011fu, c\u00fcmleler, paragraflar) metnin g\u00f6rece basit bir nitelemesini sa\u011flayabilir. Ek olarak, bir metnin kalitesini ve okunabilirli\u011fini tan\u0131mlayan \u00e7e\u015fitli karma\u015f\u0131k indisler sunar. Bu dizinler aras\u0131nda, anlat\u0131, referans uyumu, s\u00f6zdizimsel basitlik, s\u00f6zc\u00fck somutlu\u011fu ve derin bir uyum d\u00e2hil olmak \u00fczere be\u015f Coh-Metrix Metin Kolayl\u0131k Bile\u015feni bulunmaktad\u0131r (Graesser, McNamara ve Kulikowich, 2011; Jackson, Allen ve McNamara, 2016; bk. metrix.commoncoretera.com).<\/p>\n

Coh-Metrix, otomatik dil analizini herkese a\u00e7\u0131k hale getirerek dil ve s\u00f6ylem anlay\u0131\u015f\u0131m\u0131z\u0131 b\u00fcy\u00fck \u00f6l\u00e7\u00fcde etkilemi\u015ftir. Coh-Metrix birden fazla dil \u00f6l\u00e7\u00fcs\u00fc sa\u011flarken, Coh-Metrix’in \u00f6ncelikli ve \u00f6zg\u00fcn oda\u011f\u0131 metinde uyum \u00f6l\u00e7\u00fctleri sa\u011flamak olmu\u015ftur. Uyum, c\u00fcmle (yani, yerel uyum), paragraflar (yani, k\u00fcresel uyum) ve metnin geneli(\u00f6r. s\u00f6zc\u00fck \u00e7e\u015fitlili\u011fi) gibi daha b\u00fcy\u00fck b\u00f6l\u00fcmlerin, kelimeler ve anlam c\u00fcmleleriyle \u00f6rt\u00fc\u015fmesidir. Son derece yararl\u0131 olsa da Coh-Metrix’in kolayl\u0131k ve geni\u015f uyum indisleri \u00f6l\u00e7\u00fcm\u00fc ile ilgili baz\u0131 eksiklikleri vard\u0131r. \u0130lk olarak, metnin toplu i\u015flenmesine izin vermez ve kullan\u0131c\u0131n\u0131n sabit diskinde durmaz (ve bu nedenle internet ba\u011flant\u0131s\u0131na ve harici bir sunucuya ba\u011fl\u0131d\u0131r). \u0130kincisi, Coh-Metrix uyum indislerileri genel olarak k\u00fcresel uyuma (\u00f6r. bir metnin \u00e7e\u015fitli b\u00f6l\u00fcmleri aras\u0131nda semantik \u00f6rt\u00fc\u015fme) de\u011fil, yerel ve genel metin uyumuna (yani ortalama c\u00fcmle \u00e7ak\u0131\u015fmas\u0131, s\u00f6zc\u00fck \u00e7e\u015fitlili\u011fi) odaklan\u0131r. Bu nedenle, Metin Uyumunun Otomatik Analizi Arac\u0131 (MUOAA) ve Basit Do\u011fal Dil \u0130\u015fleme Arac\u0131 (BDD\u0130A) bu bo\u015fluklar\u0131 ele almak i\u00e7in geli\u015ftirilmi\u015ftir (Crossley, Allen, Kyle ve McNamara, 2014; Crossley, Kyle ve McNamara, bas\u0131m a\u015famas\u0131nda; http:\/\/www.kristopherkyle.com\/taaco.html). MUOAA yerel olarak (bilgisayara) kurulur (bir internet aray\u00fcz\u00fcne k\u0131yasla), metin dosyalar\u0131n\u0131n toplu olarak i\u015flenmesini sa\u011flar ve yerel, k\u00fcresel ve genel metin uyumu ile ilgili 150’den fazla indis i\u00e7erir. Benzer \u015fekilde, BDD\u0130A yerel olarak kurulur ve toplu metin i\u015flemeye izin verir. Bununla birlikte, BDD\u0130A, MUOAA’dan farkl\u0131d\u0131r; metinlerin bir\u00e7ok y\u00f6n\u00fcyle ilgili bilgileri hesaplamak i\u00e7in “s\u00f6zc\u00fck \u00e7antas\u0131” yakla\u015f\u0131m\u0131n\u0131 kullan\u0131r. Ek olarak, ara\u00e7 esnektir ve ara\u015ft\u0131rmac\u0131lara ek analizler sunmak i\u00e7in kendi s\u00f6zc\u00fck kategorilerini eklemelerini sa\u011flar.<\/p>\n

Serbest\u00e7e eri\u015filebilen bir DD\u0130 arac\u0131n\u0131n bir ba\u015fka \u00f6rne\u011fi, S\u00f6zc\u00fcksel Kapsaml\u0131l\u0131\u011f\u0131n Otomatik Analizi Arac\u0131d\u0131r (SKOAA; Kyle ve Crossley, 2015; http:\/\/www.kristopherkyle.com\/taales.html). SKOAA, bir metinde mevcut s\u00f6zc\u00fcksel karma\u015f\u0131kl\u0131\u011f\u0131n seviyesi hakk\u0131nda kapsaml\u0131 bilgi sa\u011flamaya odaklan\u0131r. Bu t\u00fcr bir analiz \u00f6nemlidir, \u00e7\u00fcnk\u00fc bir metnin s\u00f6zc\u00fcksel talepleri hakk\u0131nda bilgi sa\u011flamas\u0131n\u0131n yan\u0131 s\u0131ra, metnin yazar\u0131n\u0131n s\u00f6zc\u00fcksel bilgisi ile ilgili potansiyel bilgi de sa\u011flar (Kyle ve Crossley, 2015). SKOAA, bir metinde kullan\u0131lan s\u00f6zc\u00fck bilgisinin geni\u015fli\u011fini ve derinli\u011fini de\u011ferlendirmek i\u00e7in 130’un \u00fczerinde klasik ve yeni geli\u015ftirilen s\u00f6zc\u00fck indisini hesaplar. Bu ara\u00e7 h\u0131zl\u0131, g\u00fcvenilir ve \u00fccretsiz indirilebilir. SKOAA i\u00e7in al\u0131nacak \u00f6nlemler aras\u0131nda kelime s\u0131kl\u0131\u011f\u0131, kelime ve kelime ailesi \u00e7e\u015fitleri, n-gram, akademik listeler ve psikodilbilimsel unsurlar\u0131 dikkate alan kelime bilgisi indisleri bulunmaktad\u0131r (Kyle ve Crossley, 2015). Bu indisler toplu olarak, kelime se\u00e7imlerinin metindeki karma\u015f\u0131kl\u0131\u011f\u0131 hakk\u0131nda geni\u015f bilgi sa\u011flar.<\/p>\n

Dascalu, McNamara, Crossley ve Trausan-Matu (2016), ayr\u0131ca, bireysel kelimelerin \u00f6\u011frenme oran\u0131n\u0131n, \u00f6\u011frenenin dille ilgili deneyiminin bir fonksiyonu olarak hesapland\u0131\u011f\u0131 kelime karma\u015f\u0131kl\u0131\u011f\u0131n\u0131 tahmin etmek i\u00e7in hesaplanan bir model olan Maruz Kalma Ya\u015f\u0131’n\u0131 da (MKY) tan\u0131tm\u0131\u015ft\u0131r. Pearson’un kelime olgunlu\u011fu hesaplamas\u0131n\u0131n tersine (Landauer, Kireyev ve Panaccione, 2011), MKY, zaman i\u00e7inde veya daha \u00f6zel olarak art\u0131ml\u0131 olarak olu\u015fturulabilecek potansiyel \u00e7a\u011fr\u0131\u015f\u0131mlar ba\u011flam\u0131nda kelime \u00f6\u011frenimini sim\u00fcle eden yeniden \u00fcretilebilir ve \u00f6l\u00e7eklenebilir bir modeldir \u00f6zellikle art\u0131r\u0131ml\u0131 gizli Dirichlet tahsisi (Blei, Ng ve Jordan, 2003) konu modelleri. MKY indisleri, kelime s\u0131kl\u0131\u011f\u0131 ve entropi tahminlerinin yan\u0131 s\u0131ra edinim ya\u015f\u0131 ve s\u00f6zc\u00fcksel cevap gecikmelerinin insan puanlar\u0131 ile g\u00fc\u00e7l\u00fc ili\u015fkiler (kelime olgunlu\u011funun raporlanan performans\u0131n\u0131 a\u015fmaktad\u0131r) sa\u011flar.<\/p>\n

Do\u011fal Dil \u0130\u015fleme ve \u00d6\u011frenme Algoritmalar\u0131<\/h3>\n

DD\u0130, s\u00f6zc\u00fck say\u0131s\u0131, n-gram ve paragraf gibi basit tan\u0131mlay\u0131c\u0131 istatistiklerden1<\/sup><\/a> s\u00f6zc\u00fck, c\u00fcmle ve metin \u00f6zelliklerine kadar \u00e7ok say\u0131da dil y\u00fcz\u00fcn\u00fc tan\u0131mlamak i\u00e7in kullan\u0131labilir (Crossley, Allen, Kyle ve McNamara, 2014). \u015eekil 8.1’de g\u00f6sterildi\u011fi gibi, dilin bir\u00e7ok \u00f6zelli\u011fi s\u00f6zc\u00fcklerden toplan\u0131r (n-gram ve s\u00f6zc\u00fck torbalar\u0131 d\u00e2hil) ve hem g\u00f6zlemlenebilir \u00f6zellikleri analiz ederek (\u00f6r. kelime s\u0131kl\u0131klar\u0131, kelime-belge da\u011f\u0131l\u0131mlar\u0131) ve hem de metindeki gizli anlam\u0131 kullanarak elde edilebilir. (McNamara, 2011). Bilgi, kelimelerin \u00f6zellikleri, c\u00fcmleler ve metnin b\u00fct\u00fcn\u00fcyle sa\u011flan\u0131r. Bu bilgiler do\u011frusal regresyon, ay\u0131r\u0131c\u01312<\/sup><\/a> fonksiyon s\u0131n\u0131fland\u0131r\u0131c\u0131lar\u0131, Naif Bayes s\u0131n\u0131fland\u0131r\u0131c\u0131lar\u0131, destek vekt\u00f6r makineleri, yap\u0131sal ba\u011f\u0131nt\u0131 s\u0131n\u0131fland\u0131r\u0131c\u0131lar\u0131 ve karar a\u011fac\u0131 s\u0131n\u0131fland\u0131r\u0131c\u0131lar\u0131 gibi makine \u00f6\u011frenme teknikleri kullan\u0131larak analiz edilebilir. Bu teknikler \u00f6\u011frenme \u00e7\u0131kt\u0131lar\u0131n\u0131 tahmin etmek i\u00e7in kullan\u0131ld\u0131\u011f\u0131nda, daha sonra e\u011fitim teknolojileri veya uygulamalar\u0131nda kullan\u0131labilecek algoritmalar t\u00fcretilebilir. Bu uygulamalar\u0131n bir k\u0131sm\u0131n\u0131 a\u015fa\u011f\u0131daki b\u00f6l\u00fcmlerde tart\u0131\u015f\u0131yoruz.<\/p>\n

\"\"<\/p>\n

<\/a> \u015eekil 8.1. DD\u0130 kullanarak algoritmalar geli\u015ftirmek, kelimeler, c\u00fcmleler ve metnin tamam\u0131 d\u00e2hil olmak \u00fczere, metin \u00fczerindeki \u00e7e\u015fitli bilgi kaynaklar\u0131na uygulanan makine \u00f6\u011frenme tekniklerini gerektirir.<\/span><\/i><\/span><\/p>\n

YAZININ DE\u011eERLEND\u0130R\u0130LMES\u0130<\/h2>\n

DD\u0130’nin e\u011fitim alan\u0131nda kullan\u0131lmas\u0131n\u0131n en yayg\u0131n \u00f6rne\u011fi, otomatik kompozisyon puanlama (OKP) algoritmalar\u0131n\u0131n geli\u015ftirilmesidir (Allen, Jacovina ve McNamara, 2016; Dikli, 2006; Weigle, 2013; Xi, 2010). OKP sistemleri \u00e7e\u015fitli yakla\u015f\u0131mlar\u0131 kullanarak kompozisyonlar\u0131 de\u011ferlendirir. \u00d6rne\u011fin Ak\u0131ll\u0131 Kompozisyon De\u011ferlendirici (Intelligent Essay Assessor-Landauer, Laham ve Foltz, 2003) bir metnin ba\u015fka bir \u00f6l\u00e7\u00fct metne benzerli\u011fini \u00f6ncelikle \u00d6SA ya dayanarak de\u011ferlendirir. Buna kar\u015f\u0131l\u0131k, Educational Testing Service’te geli\u015ftirilen e-rater (Burstein, Chodorow ve Leacock, 2004), Vantage Learning (Rudner, Garcia ve Welch, 2006) taraf\u0131ndan geli\u015ftirilen IntelliMetric Kompozisyon Puanlama Sistemi ve Wrting Pal gibi sistemler (McNamara, Crossley ve Roscoe, 2013) DD\u0130 teknikleri ve yapay zek\u00e2n\u0131n birle\u015fimine dayan\u0131r. OKP sistemleri kompozisyon gibi yazma \u00f6rneklerini i\u015fler ve yaz\u0131n\u0131n kalitesini ve i\u00e7eri\u011fe g\u00f6re do\u011frulu\u011funu de\u011ferlendirerek yazar\u0131n g\u00f6revin taleplerini ne derece yerine getirdi\u011fini de\u011ferlendirir. OKP teknolojileri olduk\u00e7a ba\u015far\u0131l\u0131d\u0131r, genel olarak uzman insan puanlay\u0131c\u0131lar kadar kesin do\u011fruluk seviyeleri rapor eder (Attali ve Burstein, 2006; Shermis, Burstein, Higgins ve Zechner, 2010; Valenti, Neri ve Cucchiarelli, 2003; Crossley, Kyle ve McNamara), 2015).<\/p>\n

Ak\u0131ll\u0131 \u00d6\u011fretici Sistemler<\/h3>\n

DD\u0130’nin bir di\u011fer kullan\u0131m\u0131, otomatik, ak\u0131ll\u0131 ders verme teknolojileri ba\u011flam\u0131nda olmu\u015ftur. DD\u0130, \u00f6zellikle ak\u0131ll\u0131 \u00f6\u011freticilerle diyalog yoluyla etkile\u015fimde bulunanlar (\u00f6r. AutoTutor: Graesser vd., 2004) ve \u00f6\u011frenenin s\u00f6zl\u00fc cevaplar vermesini isteyen (\u00f6r. AOD\u0130ES\u00d6) bir dizi ak\u0131ll\u0131 \u00f6\u011fretici sistemine (A\u00d6S) d\u00e2hil edilmi\u015ftir. (\u00f6r. AOD\u0130ES\u00d6: McNamara, Levinstein ve Boonthum, 2004; Writing Pal: McNamara vd., 2012; Roscoe ve McNamara, 2013). Bir \u00f6\u011frenen do\u011fal dil giri\u015fi yapt\u0131\u011f\u0131nda ve yararl\u0131 geri bildirim veya makul bir cevap bekledi\u011finde, DD\u0130 bu giri\u015fi yorumlamak ve uygun geri bildirim sa\u011flamak i\u00e7in kullan\u0131labilir (McNamara vd., 2013). Do\u011fal dili girdi olarak kabul eden \u00f6zel ders sistemleri i\u00e7in (\u00f6r. metin, problemler veya bilimsel s\u00fcre\u00e7lerin s\u00f6zl\u00fc a\u00e7\u0131klamalar\u0131), \u00f6\u011frenen cevaplar\u0131 a\u00e7\u0131k u\u00e7lu ve potansiyel olarak belirsiz olabilir. \u00d6rne\u011fin, \u00f6\u011frenciye, h\u00fccre mitozunun hangi faz\u0131n\u0131n mikrot\u00fcb\u00fcllerin uzat\u0131lmas\u0131n\u0131 i\u00e7erdi\u011fi sorulabilir. Bu t\u00fcr bir soru (\u00f6r. ne veya ne zaman sorusu) k\u0131sa cevaplar veya \u00e7oktan se\u00e7meli cevaplar kullan\u0131larak cevaplanabilir ve DD\u0130 gerektirmez. Buna kar\u015f\u0131l\u0131k, anafaz s\u00fcreci a\u00e7\u0131klamak i\u00e7in bir soru yayg\u0131n olarak \u00f6\u011frenciler aras\u0131nda farkl\u0131 muhtemel cevaplar temin eder. Bu nedenle, \u00f6\u011frenenin cevab\u0131n\u0131n do\u011frulu\u011funu ve kalitesini otomatik olarak tespit etmek DD\u0130 kullan\u0131m\u0131n\u0131 gerektirir.<\/p>\n

Neden sadece \u00e7oktan se\u00e7meli kullanm\u0131yorsunuz? Bir\u00e7ok \u00f6\u011fretici sistemi tam da bunu yap\u0131yor. Bununla birlikte, \u00f6\u011frencilerin nas\u0131l ve ni\u00e7in sorular\u0131na cevap vererek bir yap\u0131 veya olgu hakk\u0131nda derinlemesine bir anlay\u0131\u015f olu\u015fturma olas\u0131l\u0131klar\u0131 daha y\u00fcksektir (\u00f6r. Johnson-Glenberg, 2007; McKeown, Beck ve Blake, 2009; Wong, 1985). Ayr\u0131ca, \u00f6\u011frencilerin bu t\u00fcr sorulara verdikleri cevaplar\u0131n onlar\u0131n anlay\u0131\u015flar\u0131n\u0131n derinli\u011fini ortaya \u00e7\u0131karmas\u0131 muhtemeldir. (Graesser ve Person, 1994; Graesser, McNamara ve VanLehn, 2005; McNamara ve Kintsch, 1996). AutoTutor, zorlu konulara dair (\u00f6r. fizik, biyoloji, bilgisayar programlamas\u0131) \u00f6\u011frencilere derinlemesine nas\u0131l ve neden soru sormalar\u0131 gerekti\u011fini s\u00f6yleyerek e\u011fitim vermeye odaklanan bir A\u00d6S’dir. AutoTutor, \u00f6\u011freneni, do\u011fru cevaplar olu\u015fturmaya y\u00f6nlendiren bir diyalogda animasyonlu bir arac\u0131yla me\u015fgul eder. Bunu, ipucu, bilgi istemleri, iddialar, d\u00fczeltmeler ve \u00f6\u011frenen sorular\u0131na cevaplar gibi \u00e7e\u015fitli diyalog hareketlerini kullanarak yapar. Bu hamleler DD\u0130 tekniklerinin bir kombinasyonu ile ger\u00e7ekle\u015ftirilir. \u00d6rne\u011fin, AutoTutor, \u00f6\u011frencilerin belirli durumlarda \u00fcretebilecekleri c\u00fcmleleri (\u00f6r. bilmiyorum; anlamad\u0131m) ve do\u011fru cevab\u0131n \u00f6nemli k\u0131s\u0131mlar\u0131n\u0131 tespit etmek i\u00e7in de\u011fi\u015fmez ifadeler kullan\u0131r. AutoTutor ayr\u0131ca, \u00f6\u011frenenin verdi\u011fi cevap ile ideal cevap aras\u0131ndaki benzerli\u011fi tespit etmek i\u00e7in \u00d6SA’y\u0131 kullan\u0131r. Sabit ifadelerin, d\u00fczenli ifadelerin veya \u00f6r\u00fcnt\u00fclerin, \u00d6SA ile \u00f6\u011frencilerin s\u00f6zel cevaplar\u0131 ve beklentileri aras\u0131ndaki ters a\u011f\u0131rl\u0131kl\u0131 s\u0131kl\u0131k s\u00f6zc\u00fck \u00f6rt\u00fc\u015fmeleri, AutoTutor’un \u00f6\u011frenenin cevab\u0131n\u0131 anlaman\u0131n benzerini \u00fcretmeye izin verir ve bu benzetilmi\u015f anlay\u0131\u015f, \u00f6\u011frenci i\u00e7in uygun bir cevap \u00fcretir. (Graesser, bas\u0131m a\u015famas\u0131nda).<\/p>\n

AOD\u0130ES\u00d6 (Aktif Okuma ve D\u00fc\u015f\u00fcnmeye Y\u00f6nelik Etkile\u015fimli Strateji E\u011fitimi), a\u00e7\u0131k u\u00e7lu cevaplara cevap vermek i\u00e7in DD\u0130 tekniklerinin birle\u015fimine dayanan bir di\u011fer A\u00d6S’dir. AOD\u0130ES\u00d6 hem DD\u0130 hem de bilgi i\u015flemsel dilbilim literat\u00fcr\u00fcnde zorlu bir g\u00f6rev olan \u00f6\u011frenenin kendi a\u00e7\u0131klamalar\u0131ndaki yorumlama sorununu ele alan ilk otomatik sistemler aras\u0131ndayd\u0131. AOD\u0130ES\u00d6, \u00f6\u011frencilerin kendini a\u00e7\u0131klamay\u0131 (\u00f6r. metni kendi kendine a\u00e7\u0131klama s\u00fcreci), ili\u015fkilendirici ve ayr\u0131nt\u0131l\u0131 \u00e7\u0131kar\u0131mlar \u00fcretme gibi anlama stratejileriyle birlikte kullanma amac\u0131yla \u00f6\u011fretim ve pratik sa\u011flayarak zorlu bilimsel metinleri kavray\u0131\u015flar\u0131n\u0131 geli\u015ftirir. AOD\u0130ES\u00d6 \u00f6\u011fretiminin uygulama a\u015famas\u0131nda, \u00f6\u011frenciler zorlu metinler i\u00e7in kendi a\u00e7\u0131klamalar\u0131n\u0131 \u00fcretirler. \u00d6\u011frencilerin AOD\u0130ES\u00d6’taki kendi a\u00e7\u0131klamalar\u0131, kelimeler hakk\u0131ndaki g\u00f6zlemlenebilir ve gizli anlamsal bilgilerin bir birle\u015fimini kullanarak kendi a\u00e7\u0131klamalar\u0131ndaki ve metindeki s\u00f6zc\u00fcklerden gelen bilgileri birle\u015ftiren bir algoritma kullan\u0131larak puanlan\u0131r, (McNamara, Boonthum, Levinstein ve Millis, 2007) ). Algoritma otomatik olarak her bir ki\u015fisel a\u00e7\u0131klama i\u00e7in 0 ile 3 aras\u0131nda bir puan atar. Daha y\u00fcksek puanlar, metin i\u00e7eri\u011fi (hem hedef c\u00fcmle hem de daha \u00f6nce okunan c\u00fcmleler) ile ilgili bilgileri i\u00e7eren ki\u015fisel a\u00e7\u0131klamalara, d\u00fc\u015f\u00fck puanlar ise ilgisiz veya k\u0131sa cevaplara verilir. Puanlama algoritmas\u0131, \u00f6\u011frencilerin hedef c\u00fcmle, \u00f6nceki metin i\u00e7eri\u011fi ve d\u00fcnya bilgisi aras\u0131nda ne \u00f6l\u00e7\u00fcde ba\u011flant\u0131 kurduklar\u0131n\u0131 yans\u0131tacak \u015fekilde tasarlanm\u0131\u015ft\u0131r. Sistem \u00e7ok \u00e7e\u015fitli metinlerdeki a\u00e7\u0131klamalar\u0131n insanlar taraf\u0131ndan geli\u015ftirilen puanlar\u0131n\u0131 ba\u015far\u0131yla e\u015fle\u015ftirir (Jackson, Guess ve McNamara, 2010; McNamara vd., 2007).<\/p>\n

Bilgisayar Destekli \u0130\u015fbirlikli \u00d6\u011frenme (BD\u0130\u00d6)<\/h3>\n

DD\u0130 teknikleri, i\u015fbirlikli \u00f6\u011frenme ortamlar\u0131nda ve \u00f6zellikle Bilgisayar Destekli \u0130\u015fbirlikli \u00d6\u011frenme (BD\u0130\u00d6) sistemlerinde ortaya \u00e7\u0131kan s\u00f6ylemde de uygulanm\u0131\u015ft\u0131r (Stahl, 2006). Bu sistemlerin bir alt k\u00fcmesi, Bakhtin (1981) taraf\u0131ndan ortaya konan, daha sonra BD\u0130\u00d6 i\u00e7in bir paradigma olarak ortaya \u00e7\u0131km\u0131\u015f olan bir kavram olan diyalojiye dayal\u0131 BD\u0130\u00d6 konu\u015fmalar\u0131n\u0131 modellemektedir (Koschmann, 1999). Dong’un tak\u0131m ileti\u015fimi tasar\u0131m\u0131, Polyphony (Trausan-Matu, Rebedea, Dragan ve Alexandru, 2007), Bilgi Alan\u0131 G\u00f6r\u00fcnt\u00fcleyici (Teplovs, 2008) ve Reader Bench’tir (Dascalu, Stavarache vd., 2015; Dascalu, Trausan-Matu, McNamara ve Dessus, 2015) en temsili yakla\u015f\u0131mlard\u0131r. ReaderBench, i\u015f birli\u011fine dayal\u0131 \u00f6\u011frenmenin yan\u0131 s\u0131ra dili anlama ile ilgili birden fazla hedefe ula\u015fmak i\u00e7in metin madencili\u011fi teknikleri, geli\u015fmi\u015f DD\u0130 ve sosyal a\u011f analizinin g\u00fcc\u00fcnden yararlanmaktad\u0131r (Dascalu, 2014). ReaderBench, kat\u0131l\u0131mc\u0131lar aras\u0131nda iletilen bilgilerin anlamsal metinsel b\u00fct\u00fcnle\u015fme yoluyla hesapland\u0131\u011f\u0131 bir Uyum A\u011f\u0131 Analizi perspektifi ile kat\u0131l\u0131m ve i\u015f birli\u011fini modellemektedir (Dascalu, Trausan-Matu, Dessus ve McNamara, 2015a). Ayr\u0131ca ReaderBench, polifonik s\u00f6ylem modeline dayanan i\u015f birli\u011fini de\u011ferlendirmek i\u00e7in otomatik bir diyalog modeli ortaya koymu\u015ftur (Trausan-Matu, Stahl ve Sarmiento, 2007). Diyaloji kuramlar\u0131na dayanarak (Bakhtin, 1981), sistem otomatik olarak sesleri veya kat\u0131l\u0131mc\u0131n\u0131n g\u00f6r\u00fc\u015f\u00fcn\u00fc, t\u00fcm konu\u015fmay\u0131 kapsayan s\u0131k\u0131 bir \u015fekilde birbirine ba\u011fl\u0131 veya anlamsal olarak ilgili kavramlar\u0131 i\u00e7eren anlamsal zincirler olarak tan\u0131mlar (Dascalu, Trausan-Matu, Dessus ve McNamara, 2015b). Bu nedenle, i\u015f birli\u011fi, farkl\u0131 kat\u0131l\u0131mc\u0131lar aras\u0131nda fikir al\u0131\u015fveri\u015fini vurgulamak i\u00e7in kullan\u0131lan ortak-olu\u015f \u00f6r\u00fcnt\u00fclerinde bilgi i\u015flemsel olarak yakalanan farkl\u0131 kat\u0131l\u0131mc\u0131 seslerinin kar\u015f\u0131l\u0131kl\u0131 canland\u0131rmas\u0131ndan ortaya \u00e7\u0131kar.<\/p>\n

Kitlesel A\u00e7\u0131k \u00c7evrimi\u00e7i Dersler (KA\u00c7D’ler)<\/h3>\n

DD\u0130’nin bir di\u011fer kullan\u0131m\u0131, \u00e7evrimi\u00e7i dersler, \u00f6zellikle de b\u00fcy\u00fck a\u00e7\u0131k \u00e7evrimi\u00e7i dersler (KA\u00c7D’ler) ba\u011flam\u0131nda olmu\u015ftur. KA\u00c7D’ler, binlerce \u00f6\u011frenciye dersleri \u00fccretsiz olarak sunmak i\u00e7in \u00e7evrimi\u00e7i platformlar\u0131 kullan\u0131r. KA\u00c7D’ler, uzaktan ve ya\u015fam boyu \u00f6\u011frenenlere eri\u015filebilirli\u011fi artt\u0131rma potansiyelleri nedeniyle \u00f6vg\u00fcyle kar\u015f\u0131lanmaktad\u0131r (Koller, Ng, Do ve Chen, 2013). Bu platformlar, tart\u0131\u015fma ak\u0131\u015flar\u0131nda ve e-postalarda \u00f6\u011frencilerin olu\u015fturdu\u011fu dillerin yan\u0131 s\u0131ra t\u0131klama ak\u0131\u015f\u0131 g\u00fcnl\u00fckleri, \u00f6devler, kurs performans\u0131 ve \u00e7ok b\u00fcy\u00fck miktarda veri sa\u011flayabilir. Bu veriler \u00f6\u011frenen tutumlar\u0131, tamamlama ve \u00f6\u011frenmeyi incelemek i\u00e7in ara\u015ft\u0131r\u0131labilir (Seaton, Bergner, Chuang, Mitros ve Pritchard, 2014; Wen, Yang ve Rose, 2014a, 2014b).<\/p>\n

KA\u00c7D’lerde \u00f6\u011frenen dilini analiz etmek i\u00e7in en yayg\u0131n DD\u0130 yakla\u015f\u0131m\u0131, duygular\u0131 analiz eden ara\u00e7lardan olmu\u015ftur. Duygu analizi, olumlu ya da olumsuz duygular\u0131n dilini ya da motivasyon, anla\u015fma, bili\u015fsel mekanizmalar ya da kat\u0131l\u0131mla ilgili kelimeleri inceler (Chaturvedi, Goldwasser ve Daume, 2014; Elouazizi, 2014; Moon, Potdar ve Martin, 2014; Wen vd., 2014a, 2014b). \u00d6rne\u011fin, Moon vd. (2014), \u00f6\u011frenen liderlerini tan\u0131mlamak i\u00e7in kat\u0131l\u0131mc\u0131lar aras\u0131nda duygu terimleri ve anlamsal benzerlikler kullanm\u0131\u015ft\u0131r. Bak\u0131\u015f a\u00e7\u0131s\u0131na ili\u015fkin dil indislerinin (\u00f6r. san\u0131r\u0131m, bence b\u00fcy\u00fck ihtimalle, muhtemelen) kursun d\u00fc\u015f\u00fck kat\u0131l\u0131m d\u00fczeyleriyle ili\u015fkili oldu\u011funu g\u00f6stermi\u015ftir. Wen ve meslekta\u015flar\u0131, (2014a, 2014b), \u00f6\u011frencilerin tart\u0131\u015fma zamirlerini ve tart\u0131\u015fma forumlar\u0131ndaki motivasyonla ilgili s\u00f6zc\u00fckleri kullanmalar\u0131n\u0131n, dersten ayr\u0131lma riskinin daha d\u00fc\u015f\u00fck olaca\u011f\u0131n\u0131 \u00f6ng\u00f6rd\u00fc\u011f\u00fcn\u00fc bulmu\u015flard\u0131r.<\/p>\n

Benzer \u015fekilde, Crossley, McNamara vd. (2015), \u00f6\u011frencilerin dilini, KA\u00c7D tart\u0131\u015fma forumunda, e\u011fitsel veri madencili\u011fi konusunu kapsayan bir kursta incelemek i\u00e7in \u00e7ok say\u0131da dil \u00f6zelli\u011fi kullanm\u0131\u015ft\u0131r (Baker vd., bas\u0131m a\u015famas\u0131nda). Crossley, McNamara vd. (2015) KA\u00c7D tart\u0131\u015fma forumlar\u0131nda kat\u0131lan 320 \u00f6\u011frencinin (\u00f6r. g\u00f6nderilen 49 kelime) tamamlama oranlar\u0131n\u0131 ba\u015far\u0131 ile (%70 do\u011frulukla) tahmin etmi\u015ftir. Kursta bitirme sertifikas\u0131 alma olas\u0131l\u0131\u011f\u0131 daha y\u00fcksek olan \u00f6\u011frenciler genellikle daha karma\u015f\u0131k bir dil kulland\u0131lar. \u00d6rne\u011fin, onlar\u0131n mesajlar\u0131 daha anla\u015f\u0131l\u0131r ve tutarl\u0131, daha s\u0131k ve belirli bir kelime kullan\u0131lm\u0131\u015f ve daha genel yazma niteli\u011fine sahipti. \u0130lgin\u00e7tir ki, duyu\u015f ile ilgili indisler tamamlanma oranlar\u0131n\u0131 \u00f6ng\u00f6rm\u00fcyordu.<\/p>\n

Toplu olarak, bu ara\u015ft\u0131rma DD\u0130’nin KA\u00c7D’lerin \u00f6\u011fretim g\u00f6revlisi ile \u00f6\u011frenciler aras\u0131ndaki ve \u00f6\u011frencilerin kendi aras\u0131ndaki ileti\u015fimi ba\u011flam\u0131nda g\u00fc\u00e7l\u00fc bir ba\u015far\u0131 g\u00f6stergesi olabilece\u011fine dair umut verici kan\u0131tlar sunar ve bu \u00f6zellikle de uzaktan kurslar i\u00e7in \u00e7ok \u00f6nemlidir. Ayr\u0131ca, bu ileti\u015fim daha sonra \u00f6\u011frenen performans\u0131n\u0131n de\u011ferlendirme formlar\u0131 olarak da kullan\u0131labilir. Bu nedenle, KA\u00c7D’lerin \u00f6\u011frenen kat\u0131l\u0131m\u0131n\u0131 ve potansiyel ba\u015far\u0131s\u0131n\u0131 daha iyi izlemek i\u00e7in tart\u0131\u015fma forumlar\u0131n\u0131 i\u00e7ermesi gerekti\u011fi a\u00e7\u0131k g\u00f6r\u00fcnmektedir. \u00d6\u011frencilerin kulland\u0131\u011f\u0131 dil, kursu tamamlama olas\u0131l\u0131\u011f\u0131 daha d\u00fc\u015f\u00fck olan \u00f6\u011frencileri belirlemek ve bu \u00f6\u011frencilere e-posta g\u00f6ndermek, i\u00e7erik \u00f6nermek veya \u00f6zel ders \u00f6nermek gibi m\u00fcdahaleler hedeflemek i\u00e7in de kullan\u0131labilir. Dil anlay\u0131\u015f\u0131n\u0131 otomatikle\u015ftirmek ve b\u00f6ylece bu kurslardaki dil ve sosyal etkile\u015fimler hakk\u0131nda bilgi vermek, KA\u00c7D’lerde hem \u00f6\u011frenmeyi hem de etkile\u015fimi geli\u015ftirmeye yard\u0131mc\u0131 olacakt\u0131r.<\/p>\n

DD\u0130’nin G\u00dcC\u00dc<\/h2>\n

DD\u0130, \u00f6ncelikle dilin her yerde olmas\u0131 ve ayn\u0131 zamanda dili analiz etme ara\u00e7lar\u0131n\u0131n dilin neredeyse her y\u00f6n\u00fcyle ilgili g\u00f6stergeler sa\u011flamas\u0131 nedeniyle olduk\u00e7a g\u00fc\u00e7l\u00fcd\u00fcr (Crossley, 2013). DD\u0130 kullan\u0131lan belirli kelimeleri, kelime gruplar\u0131n\u0131 ve kelimeler aras\u0131ndaki ve daha b\u00fcy\u00fck metin g\u00f6vdeleri aras\u0131ndaki ili\u015fkilerin g\u00fcc\u00fcn\u00fc tespit edebilir. Ayr\u0131ca, metnin s\u0131kl\u0131\u011f\u0131, somutlu\u011fu veya anlaml\u0131l\u0131\u011f\u0131, c\u00fcmlelerin karma\u015f\u0131kl\u0131\u011f\u0131 ve metnin uyum ve t\u00fcr gibi \u00e7e\u015fitli y\u00f6nleri gibi metnin \u00f6zelliklerini de alg\u0131layabilir. Kelimeler ve \u00f6zellikleri, \u00e7e\u015fitli yap\u0131lar i\u00e7in vekil g\u00f6revi g\u00f6r\u00fcr. \u00d6rne\u011fin, bir metindeki kelimelerin s\u0131kl\u0131\u011f\u0131, metni anlamak i\u00e7in gerekli olabilecek bilgiyi tahmin etmede bir vekil olarak hizmet eder. Bir metnin birle\u015ftirilmesi, bir metindeki bo\u015fluklar\u0131 doldurmak i\u00e7in gerekli olan bilginin bir tahminini verir.<\/p>\n

DD\u0130, \u00e7ok \u00e7e\u015fitli ba\u015fka yap\u0131lar\u0131 tan\u0131mlamak i\u00e7in kullan\u0131lm\u0131\u015ft\u0131r. \u00d6rne\u011fin, Crossley ve McNamara (2012), ikinci dil (D2) yazarlar\u0131n\u0131n makalelerinin dil \u00f6zelliklerinin, bu yazarlar\u0131n ana dilini tahmin edebilece\u011fini g\u00f6stermi\u015ftir. Varner, Roscoe ve McNamara (2013) hem Coh-Metrix hem de DSKS taraf\u0131ndan sa\u011flanan g\u00f6stergeleri kullanarak \u00f6\u011frencilerin ve \u00f6\u011fretmenlerin kompozisyon kalitesi puanlar\u0131ndaki farkl\u0131l\u0131klar\u0131 incelemi\u015ftir. Louwerse, McCarthy, McNamara ve Graesser (2004), konu\u015fulan veyaz\u0131l\u0131 \u0130ngilizce \u00f6rnekleri aras\u0131ndaki farklar\u0131 belirlemek i\u00e7in DD\u0130 tekniklerini kulland\u0131. McCarthy, Briner, Rus ve McNamara (2007) Coh-Metrix’in giri\u015f, y\u00f6ntem, sonu\u00e7 ve tart\u0131\u015fma gibi tipik bilimsel metinlerdeki b\u00f6l\u00fcmleri farkl\u0131la\u015ft\u0131rabildi\u011fini g\u00f6stermi\u015ftir. Ek olarak, Crossley, Louwerse, McCarthy ve McNamara’n\u0131n (2007) ikinci dil \u00f6\u011frenenlerin metinlerinin incelemeleri, ikinci dil \u00f6\u011frenme ama\u00e7lar\u0131 i\u00e7in edinilmi\u015f (veya otantik) ile adapte edilmi\u015f (veya basitle\u015ftirilmi\u015f) metinler aras\u0131nda \u00e7ok \u00e7e\u015fitli yap\u0131sal ve s\u00f6zc\u00fcksel farkl\u0131l\u0131klar ortaya koydu. Son olarak, DD\u0130 aldat\u0131c\u0131l\u0131\u011f\u0131 tespit etmek i\u00e7in de kullan\u0131lm\u0131\u015ft\u0131r. Duran, Hall, McCarthy ve McNamara (2010), bir ki\u015finin aldat\u0131c\u0131 oldu\u011fu konu\u015fma diyaloglar\u0131 ile samimi oldu\u011fu konu\u015fmalar aras\u0131nda dilin hangi \u00f6zelliklerinin ay\u0131r\u0131c\u0131 oldu\u011funu ara\u015ft\u0131rd\u0131.<\/p>\n

DD\u0130’nin kullanman\u0131n olas\u0131 sak\u0131ncalar\u0131 oldu\u011funa dikkat etmek \u00f6nemlidir. \u00d6rne\u011fin, belirli DD\u0130 teknikleri, kelime say\u0131mlar\u0131n\u0131 veya “s\u00f6zc\u00fck \u00e7antas\u0131” yakla\u015f\u0131mlar\u0131n\u0131 kullanan diyalogun basitle\u015ftirilmi\u015f temsillerine dayan\u0131r. \u00d6SA vekt\u00f6r uzaylar\u0131, gizli Dirichlet tahsisi konu da\u011f\u0131l\u0131mlar\u0131 (GDT; Blei vd., 2003) ve sinir a\u011flar\u0131na dayal\u0131 word2vec modelleri d\u00e2hil olmak \u00fczere en \u00f6nemli ve en yayg\u0131n kullan\u0131lan DD\u0130 kelime g\u00f6sterimlerinin (Mikolov, Chen, Corrado ve Dean, 2013) hepsi, kelime s\u0131ras\u0131n\u0131n dikkate al\u0131nmad\u0131\u011f\u0131 \u201cs\u00f6zc\u00fck \u00e7antas\u0131\u201d varsay\u0131m\u0131na tabidir. Ek olarak, bir\u00e7ok DD\u0130 analizi, konu\u015fmac\u0131n\u0131n niyetleri veya pragmatik y\u00f6nleri gibi ba\u011flamlar\u0131 g\u00f6rmezden gelir. Benzer \u015fekilde, DD\u0130 analizleri genellikle belirli \u015firketlerle ve durumlarla s\u0131n\u0131rl\u0131d\u0131r ve di\u011fer ba\u011flamlara genellenememektedir. Bu (ve di\u011fer) uyar\u0131larla bile DD\u0130 son derece g\u00fc\u00e7l\u00fcd\u00fcr. DD\u0130 ara\u00e7lar\u0131ndan edinilebilecek geni\u015f bilgi kaynaklar\u0131 nedeniyle ve kulland\u0131\u011f\u0131m\u0131z dil, d\u00fc\u015f\u00fcnceler ve niyetleri temsil eden bir uzant\u0131 veya haricile\u015ftirme olabilece\u011finden DD\u0130 bireyler, yetenekleri, duygular\u0131, niyetleri ve sosyal etkile\u015fimleri hakk\u0131nda bilgi sa\u011flayabilir. \u00d6\u011frenme analiti\u011fi ba\u011flam\u0131nda, \u00f6\u011frenme s\u00fcre\u00e7lerini ve \u00f6\u011freneni otomatik olarak anlama yolunda bir ara\u00e7t\u0131r.<\/p>\n

B\u00fcy\u00fck Resim<\/h3>\n

DD\u0130, ara\u015ft\u0131rmac\u0131lar\u0131n dil ve onun ya\u015fam\u0131n \u00e7e\u015fitli y\u00f6nlerinde potansiyel olarak oynad\u0131\u011f\u0131 rolleri daha iyi anlayabilmelerini sa\u011flayan dil analizini otomatikle\u015ftiren teknikler sunar. DD\u0130, \u00f6\u011freneni sorulara, a\u00e7\u0131klamalara ve kompozisyon cevaplar\u0131 i\u00e7inde bir dil olu\u015fturmas\u0131 i\u00e7in y\u00f6nlendiren ak\u0131ll\u0131 \u00f6\u011fretici sistemlerindeki geri bildirim sistemlerini bilgilendirir. DD\u0130 ders tabanl\u0131 sistemler dil i\u00e7inde ak\u0131ll\u0131 sim\u00fclasyonu i\u00e7in bir ara\u00e7 sa\u011flar. DD\u0130 ayr\u0131ca \u00e7evrimi\u00e7i tart\u0131\u015fma forumlar\u0131 ba\u011flam\u0131nda da bilgilendiricidir. \u00d6\u011frencilerin iyi performans g\u00f6sterme veya kursu tamamlama ihtimalini yordayarak, \u00f6\u011frenci tutumlar\u0131, motivasyonu ve dilin kalitesi hakk\u0131nda bilgi sa\u011flar.<\/p>\n

\u00d6\u011frenme analiti\u011finin bir amac\u0131, daha etkili bir \u00f6\u011fretim sa\u011flamak i\u00e7in \u00f6\u011frencilerin \u00f6zelliklerini ve becerilerini modellemektir (Allen ve McNamara, 2015). Bu verileri \u00f6zellikle, \u00e7e\u015fitli ama\u00e7lar i\u00e7in kullanabiliriz: Performansla ilgili otomatik geri bildirim sa\u011flama, \u00f6\u011frenme s\u0131ras\u0131nda m\u00fcdahale etme, y\u00f6nlendirme veya bili\u015fsel deste\u011fi sa\u011flama, \u00f6zel ders \u00f6nerme, analiz verilerinden elde edilen bilgilerin \u00f6\u011frenimi geli\u015ftirece\u011fi varsay\u0131m\u0131yla izleme, \u00f6\u011frenmeyi ki\u015fiselle\u015ftirme vb. Bu ama\u00e7la, ara\u015ft\u0131rmac\u0131lar giderek daha b\u00fcy\u00fck, karma\u015f\u0131k veri kaynaklar\u0131na (yani, b\u00fcy\u00fck veriye) d\u00f6nmekte ve \u00e7e\u015fitli veri t\u00fcrleri ve analitik tekniklerin bile\u015fimini kullanmaktad\u0131r. Bu \u00e7aba i\u00e7in DD\u0130 \u00e7ok \u00f6nemlidir, \u00e7\u00fcnk\u00fc \u00f6nerilen teknikler, \u00e7e\u015fitli ba\u011flamlarda anlama d\u00fczeyini tahmin etme ve de\u011ferlendirme yoluyla \u00f6\u011frenenin \u00f6\u011frenmesini geli\u015ftirmeye yard\u0131mc\u0131 olur. Ancak DD\u0130 bulmacan\u0131n yaln\u0131zca bir par\u00e7as\u0131d\u0131r.<\/p>\n

\"\"<\/p>\n

<\/a> \u015eekil 8.2. E\u011fitsel sonu\u00e7lar\u0131n\u0131n \u00f6ng\u00f6r\u00fclmesi, birden fazla veri kayna\u011f\u0131n\u0131n birle\u015ftirilmesini gerektirecektir.<\/i><\/span><\/p>\n

\u015eekil 8.2’de g\u00f6sterildi\u011fi gibi, \u00f6\u011frenen \u00e7\u0131kt\u0131lar\u0131n\u0131n eksiksiz ve son derece kestirimsel bir anlay\u0131\u015f\u0131n\u0131n geli\u015ftirilmesi, \u00e7ok say\u0131da bilgi kayna\u011f\u0131n\u0131 ve veri analizine y\u00f6nelik \u00e7e\u015fitli yakla\u015f\u0131mlar\u0131 gerektirir. \u00d6\u011frenme, \u00e7oklu katmanlara ve \u00e7oklu zaman \u00f6l\u00e7eklerine sahip karma\u015f\u0131k bir s\u00fcre\u00e7tir. \u00d6\u011frenme s\u00fcrecini anlamak i\u00e7in herhangi bir kayna\u011fa veya veri t\u00fcr\u00fcne g\u00fcvenmek, \u00f6zellikle \u015fu anda \u00e7ok say\u0131da otomatik bilgi kayna\u011f\u0131 mevcut oldu\u011funda, uza\u011f\u0131 g\u00f6remeyen bir yakla\u015f\u0131md\u0131r. DD\u0130, nihai olarak arad\u0131\u011f\u0131m\u0131z b\u00fcy\u00fck resmin ayr\u0131lmaz bir par\u00e7as\u0131 olarak giderek daha fazla tan\u0131nan bir veri kayna\u011f\u0131d\u0131r. Tam bir \u00f6\u011frenme anlay\u0131\u015f\u0131 geli\u015ftirmek, birden fazla veri kayna\u011f\u0131n\u0131n birle\u015ftirilmesini gerektirecektir.<\/p>\n

TE\u015eEKK\u00dcR B\u00d6L\u00dcM\u00dc<\/h2>\n

Bu b\u00f6l\u00fcm\u00fcn bir k\u0131sm\u0131 E\u011fitim Bilimleri Enstit\u00fcs\u00fc (EBE R305A120707, R305A130124), Ulusal Bilim Vakf\u0131 (UBV DRL-1319645, DRL-1418352, DRL-1418378, DRL-1417997) ve Denizcilik Ara\u015ft\u0131rma B\u00fcrosu taraf\u0131ndan desteklenmi\u015ftir. Ara\u015ft\u0131rma (DAB N000141410343). Bu yaz\u0131da dile getirilen herhangi bir g\u00f6r\u00fc\u015f, bulgu ve \u00e7\u0131kar\u0131mlar veya tavsiye yazarlara aittir ve IES, UBV veya DAB’nin g\u00f6r\u00fc\u015flerini yans\u0131tmak zorunda de\u011fildir. DD\u0130 konusundaki ara\u015ft\u0131rmalar\u0131m\u0131za y\u0131llar boyunca katk\u0131da bulunan bir\u00e7ok \u00f6\u011frenciye, doktora sonras\u0131 ara\u015ft\u0131rmac\u0131ya ve fak\u00fclteye minnettar\u0131z.<\/p>\n

KAYNAK\u00c7A<\/h2>\n

Allen, L. K., Jacovina, M. E., & McNamara, D. S. (2016). Computer-based writing instruction. In C. A. MacArthur, S. Graham, & J. Fitzgerald (Eds.), <\/span>Handbook of writing research<\/i><\/span>, 2nd ed. (S. 316-329). New York: The Guilford Press. <\/span><\/span><\/p>\n

Allen, L. K., & McNamara, D. S. (2015). You are your words: Modeling students’ vocabulary knowledge with natural language processing. In O. C. Santos, J. G. Boticario, C. Romero, M. Pechenizkiy, A. Merceron, P. Mitros, J. M. Luna, C. Mihaescu, P. Moreno, A. Hershkovitz, S. Ventura, & M. Desmarais (Eds.), Proceedings of the 8th International Conference on Educational Data Mining <\/i>(EDM 2015) 26\u201329 June 2015, Madrid, Spain (pp. 258\u2013265). International Educational Data Mining Society. <\/span><\/p>\n

Attali, Y., & Burstein, J. (2006). Automated essay scoring with e-rater\u00ae V. 2. The Journal of Technology, Learning and Assessment, 4<\/i>(2). doi:10.1002\/j.2333-8504.2004.tb01972.x <\/span><\/p>\n

Baker, R., Wang, E., Paquette, L., Aleven, V., Popescu, O., Sewall, J., Rose, C., Tomar, G., Ferschke, O., Hollands, F., Zhang, J., Cennamo, M., Ogden, S., Condit, T., Diaz, J., Crossley, S., McNamara, D., Comer, D., Lynch, C., Brown, R., Barnes, T., & Bergner, Y. (in press). A MOOC on educational data mining. In. S. ElAtia, O. Za\u00efane, & D. Ipperciel (Eds.). Data Mining and Learning Analytics in Educational Research<\/i>. Wiley & Blackwell. <\/span><\/p>\n

Bakhtin, M.M. (1981). The dialogic imagination: Four essays <\/i>(C. Emerson & M. Holquist, Trans.). Austin, TX: University of Texas Press. <\/span><\/p>\n

Blei, D. M., Ng, A. Y., & Jordan, M. I. (2003). Latent Dirichlet allocation. Journal of Machine Learning Research, 3<\/i>(4\u20135), 993\u20131022. <\/span><\/p>\n

Burstein, J., Chodorow, M., & Leacock, C. (2004). Automated essay evaluation: The Criterion online writing service. Ai Magazine, 25<\/i>(3), 27.<\/span><\/p>\n

Chaturvedi, S., Goldwasser, D., & Daum\u00e9 III, H. (2014). Predicting instructor’s intervention in MOOC forums. In D. Marcu, K. Toutanova, & H. W. Baidu (Eds.), Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics <\/i>(pp. 1501\u20131511). Baltimore, MD. <\/span><\/p>\n

Crossley, S. A. (2013). Advancing research in second language writing through computational tools and machine learning techniques: A research agenda. Language Teaching, 46<\/i>(2), 256\u2013271. <\/span><\/p>\n

Crossley, S. A., Allen, L. K., Kyle, K., & McNamara, D. S. (2014). Analyzing discourse processing using a simple natural language processing tool (SiNLP). Discourse Processes, 51<\/i>, 511\u2013534. <\/span><\/p>\n

Crossley, S. A., Kyle, K., & McNamara, D. S. (2015). To aggregate or not? Linguistic features in automatic essay scoring and feedback systems. Journal of Writing Assessment, 8<\/i>(1). http:\/\/www.journalofwritingassessment.org\/article.php?article=80 <\/span><\/p>\n

Crossley, S. A. Kyle, K., & McNamara, D. S. (in press). Tool for the automatic analysis of text cohesion (TAACO): Automatic assessment of local, global, and text cohesion. Behavior Research Methods<\/i>. <\/span><\/p>\n

Crossley, S. A., & Louwerse, M. (2007). Multi-dimensional register classification using bigrams. International Journal of Corpus Linguistics, 12<\/i>(4), 453\u2013478. <\/span><\/p>\n

Crossley, S. A., Louwerse, M., McCarthy, P. M., & McNamara, D. S. (2007). A linguistic analysis of simplified and authentic texts. Modern Language Journal, 91<\/i>, 15\u201330. <\/span><\/p>\n

Crossley, S. A., & McNamara, D. S. (2012). Interlanguage talk: A computational analysis of non-native speakers’ lexical production and exposure. In P. M. McCarthy & C. Boonthum-Denecke (Eds.), Applied natural language processing and content analysis: Identification, investigation, and resolution <\/i>(pp. 425\u2013437). Hershey, PA: IGI Global. <\/span><\/p>\n

Crossley, S. A., McNamara, D. S., Baker, R., Wang, Y., Paquette, L., Barnes, T., & Bergner, Y. (2015). Language to completion: Success in an educational data mining massive open online class. In O. C. Santos, J. G. Boticario, C. Romero, M. Pechenizkiy, A. Merceron, P. Mitros, J. M. Luna, C. Mihaescu, P. Moreno, A. Hershkovitz, S. Ventura, & M. Desmarais (Eds.), Proceedings of the 8th International Conference on Educational Data Mining <\/i>(EDM 2015), 26\u201329 June 2015, Madrid, Spain (pp. 388\u2013391). International Educational Data Mining Society. <\/span><\/p>\n

Dascalu, M. (2014). Analyzing discourse and text complexity for learning and collaborating. Studies in Computational Intelligence <\/i>(Vol. 534). Switzerland: Springer. <\/span><\/p>\n

Dascalu, M., Stavarache, L. L., Dessus, P., Trausan-Matu, S., McNamara, D. S., & Bianco, M. (2015). ReaderBench: The learning companion. In A. Mitrovic, F. Verdejo, C. Conati, & N. Heffernan (Eds.), Proceedings of the 17th International Conference on Artificial Intelligence in Education <\/i>(AIED\u201915), 22\u201326 June 2015, Madrid, Spain (pp. 915\u2013916). Springer. <\/span><\/p>\n

Dascalu, M., Trausan-Matu, S., Dessus, P., & McNamara, D. S. (2015a). Discourse cohesion: A signature of collaboration. In P. Blikstein, A. Merceron, & G. Siemens (Eds.), Proceedings of the 5th International Learning Analytics & Knowledge Conference <\/i>(LAK\u201915), 16\u201320 March, Poughkeepsie, NY, USA (pp. 350\u2013354). New York: ACM. <\/span><\/p>\n

Dascalu, M., Trausan-Matu, S., Dessus, P., & McNamara, D. S. (2015b). Dialogism: A framework for CSCL and a signature of collaboration. In O. Lindwall, P. H\u00e4kkinen, T. Koschmann, P. Tchounikine, & S. Ludvigsen (Eds.), Proceedings of the 11th International Conference on Computer-Supported Collaborative Learning <\/i>(CSCL 2015), 7\u201311 June 2015, Gothenburg, Sweden (pp. 86\u201393). International Society of the Learning Sciences. <\/span><\/p>\n

Dascalu, M., Trausan-Matu, S., McNamara, D. S., & Dessus, P. (2015). ReaderBench: Automated evaluation of collaboration based on cohesion and dialogism. International Journal of Computer-Supported Collaborative Learning, 10<\/i>(4), 395\u2013423. <\/span><\/p>\n

Dascalu, M., McNamara, D. S., Crossley, S. A., & Trausan-Matu, S. (2016). Age of exposure: A model of word learning. Proceedings of the 30th Conference on Artificial Intelligence <\/i>(AAAI-16), 12\u201317 February 2016, Phoenix, Arizona, USA (pp. 2928\u20132934). Palo Alto, CA: AAAI Press. <\/span><\/p>\n

Dikli, S. (2006). An overview of automated scoring of essays. The Journal of Technology, Learning and Assessment, 5<\/i>(1). http:\/\/files.eric.ed.gov\/fulltext\/EJ843855.pdf<\/span><\/p>\n

Dong, A. (2005). The latent semantic approach to studying design team communication. Design Studies, 26<\/i>(5), 445\u2013461. <\/span><\/p>\n

Duran, N. D., Hall, C., McCarthy, P. M., & McNamara, D. S. (2010). The linguistic correlates of conversational deception: Comparing natural language processing technologies. Applied Psycholinguistics, 31<\/i>(3), 439\u2013462. <\/span><\/p>\n

Elouazizi, N. (2014). Point-of-view mining and cognitive presence in MOOCs: A (computational) linguistics perspective. EMNLP 2014<\/i>, 32. http:\/\/www.aclweb.org\/anthology\/W14-4105 <\/span><\/p>\n

Graesser, A. C. (in press). Conversations with AutoTutor help students learn. International Journal of Artificial Intelligence in Education<\/i>. <\/span><\/p>\n

Graesser, A. C., Lu, S., Jackson, G. T., Mitchell, H. H., Ventura, M., Olney, A., & Louwerse, M. M. (2004). AutoTutor: A tutor with dialogue in natural language. Behavior Research Methods, Instruments, & Computers, 36<\/i>(2), 180\u2013192. <\/span><\/p>\n

Graesser, A. C., McNamara, D. S., & Kulikowich, J. M. (2011). Coh Metrix: Providing multilevel analyses of text characteristics. Educational Researcher, 40<\/i>, 223\u2013234. <\/span><\/p>\n

Graesser, A. C., McNamara, D. S., & VanLehn, K. (2005). Scaffolding deep comprehension strategies through Point & Query, AutoTutor, and iSTART. Educational Psychologist, 40<\/i>, 225\u2013234. <\/span><\/p>\n

Graesser, A. C., & Person, N. K. (1994). Question asking during tutoring. American Educational Research Journal, 31<\/i>(1), 104\u2013137. <\/span><\/p>\n

Jackson, G. T., Allen, L. K., & McNamara, D. S. (2016). Common Core TERA: Text Ease and Readability Assessor. In S. A. Crossley & D. S. McNamara (Eds.), Adaptive educational technologies for literacy instruction<\/i>. New York: Taylor & Francis, Routledge. <\/span><\/p>\n

Jackson, G. T., Guess, R. H., & McNamara, D. S. (2010). Assessing cognitively complex strategy use in an untrained domain. Topics in Cognitive Science, 2<\/i>, 127\u2013137. <\/span><\/p>\n

Jarvis, S., Bestgen, Y., Crossley, S. A., Granger, S., Paquot, M., Thewissen, J., & McNamara, D. S. (2012). The comparative and combined contributions of n-grams, Coh-Metrix indices and error types in the L1 classification of learner texts. In S. Jarvis & S. A. Crossley (Eds.), Approaching language transfer through text classification: Explorations in the detection-based approach <\/i>(pp. 154\u2013177). Bristol, UK: Multilingual Matters. <\/span><\/p>\n

Johnson-Glenberg, M. C. (2007). Web-based reading comprehension instruction: Three studies of 3D-readers. In D. McNamara (Ed.), Reading comprehension strategies: Theory, interventions, and technologies <\/i>(pp. 293\u2013324). Mahwah, NJ: Lawrence Erlbaum Publishers. <\/span><\/p>\n

Jurafsky, D., & Martin, J. H. (2000). Speech and language processing: An introduction to natural language processing, computational linguistics and speech recognition. Upper Saddle River, NJ: Prentice Hall. <\/span><\/p>\n

Jurafsky, D., & Martin, J. H. (2008). Speech and language processing: An introduction to natural language processing, computational linguistics and speech recognition, 2nd ed. Upper Saddle River, NJ: Prentice Hall. <\/span><\/p>\n

Koller, D., Ng, A., Do, C., & Chen, Z. (2013). Retention and intention in massive open online courses: In depth. Educause Review, 48<\/i>(3), 62\u201363. <\/span><\/p>\n

Koschmann, T. (1999). Toward a dialogic theory of learning: Bakhtin’s contribution to understanding learning in settings of collaboration. In C. M. Hoadley & J. Roschelle (Eds.), Proceedings of the 1999 Conference on Computer Support for Collaborative Learning <\/i>(CSCL\u201999), 12\u201315 December 1999, Palo Alto, California (pp. 308\u2013313). International Society of the Learning Sciences. <\/span><\/p>\n

Kyle, K., & Crossley, S. A. (2015). Automatically assessing lexical sophistication: Indices, tools, findings, and application. TESOL Quarterly, 49<\/i>(4), 757\u2013786. <\/span><\/p>\n

Landauer, T. K., & Dumais, S. T. (1997). A solution to Plato’s problem: The latent semantic analysis theory of acquisition, induction, and representation of knowledge. Psychological Review, 104<\/i>(2), 211\u2013240. <\/span><\/p>\n

Landauer, T. K., Laham, D., & Foltz, P. W. (2003). Automated scoring and annotation of essays with the Intelligent Essay Assessor. In M. D. Shermis & J. Burstein (Eds.), Automated essay scoring: A cross-disciplinary perspective <\/i>(pp. 87\u2013112). Mahwah, NJ: Lawrence Erlbaum.<\/span><\/p>\n

Landauer, T., McNamara, D. S., Dennis, S., & Kintsch, W. (Eds.). (2007). Handbook of latent semantic analysis<\/i>. Mahwah, NJ: Lawrence Erlbaum. <\/span><\/p>\n

Landauer, T. K., Kireyev, K., & Panaccione, C. (2011). Word maturity: A new metric for word knowledge. Scientific Studies of Reading, 15<\/i>(1), 92\u2013108. <\/span><\/p>\n

Louwerse, M. M., McCarthy, P. M., McNamara, D. S., & Graesser, A. C. (2004). Variation in language and cohesion across written and spoken registers. In K. Forbus, D. Gentner, & T. Regier (Eds.), Proceedings of the 26th Annual Conference of the Cognitive Science Society <\/i>(CogSci 2004), 4\u20137 August 2004, Chicago, IL, USA (pp. 843\u2013848). Mahwah, NJ: Lawrence Erlbaum. <\/span><\/p>\n

McCarthy, P. M., Briner, S. W., Rus, V., & McNamara, D. S. (2007). Textual signatures: Identifying text-types using latent semantic analysis to measure the cohesion of text structures. In A. Kao & S. Poteet (Eds.), Natural language processing and text mining <\/i>(pp. 107\u2013122). London: Springer-Verlag UK. <\/span><\/p>\n

McKeown, M. G., Beck, I. L., & Blake, R. G. K. (2009). Rethinking reading comprehension instruction: A comparison of instruction for strategies and content approaches. Reading Research Quarterly, 44<\/i>, 218\u2013253. <\/span><\/p>\n

McNamara, D. S. (2011). Computational methods to extract meaning from text and advance theories of human cognition. Topics in Cognitive Science, 2<\/i>, 1\u201315. <\/span><\/p>\n

McNamara, D. S., Boonthum, C., Levinstein, I. B., & Millis, K. (2007). Evaluating self-explanations in iSTART: Comparing word-based and LSA algorithms. In T. Landauer, D. S. McNamara, S. Dennis, & W. Kintsch (Eds.), Handbook of latent semantic analysis <\/i>(pp. 227\u2013241). Mahwah, NJ: Lawrence Erlbaum. <\/span><\/p>\n

McNamara, D. S., Crossley, S. A., & Roscoe, R. D. (2013). Natural language processing in an intelligent writing strategy tutoring system. Behavior Research Methods, 45<\/i>, 499\u2013515. <\/span><\/p>\n

McNamara, D. S., Graesser, A. C., McCarthy, P., & Cai, Z. (2014). Automated evaluation of text and discourse with Coh-Metrix<\/i>. Cambridge, UK: Cambridge University Press. <\/span><\/p>\n

McNamara, D. S., & Kintsch, W. (1996). Learning from text: Effects of prior knowledge and text coherence. Discourse Processes, 22<\/i>, 247\u2013288. <\/span><\/p>\n

McNamara, D. S., Levinstein, I. B., & Boonthum, C. (2004). iSTART: Interactive strategy trainer for active reading and thinking. Behavioral Research Methods, Instruments, & Computers, 36<\/i>, 222\u2013233. <\/span><\/p>\n

McNamara, D. S., Ozuru, Y., Graesser, A. C., & Louwerse, M. (2006). Validating Coh-Metrix. In R. Sun & N. Miyake (Eds.), Proceedings of the 28th Annual Conference of the Cognitive Science Society <\/i>(CogSci 2006), 26\u201329 July 2006, Vancouver, British Columbia, Canada (pp. 573\u2013578). Austin, TX: Cognitive Science Society. <\/span><\/p>\n

McNamara, D. S., Raine, R., Roscoe, R., Crossley, S. A, Jackson, G. T., Dai, J., Cai, Z., Renner, A., Brandon, R., Weston, J., Dempsey, K., Carney, D., Sullivan, S., Kim, L., Rus, V., Floyd, R., McCarthy, P. M., & Graesser, A. C. (2012). The Writing-Pal: Natural language algorithms to support intelligent tutoring on writing strategies. In P. M. McCarthy & C. Boonthum-Denecke (Eds.), Applied natural language processing and content analysis: Identification, investigation, and resolution <\/i>(pp. 298\u2013311). Hershey, PA: IGI Global. <\/span><\/p>\n

Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word representation in vector space. In Workshop at ICLR<\/i>. Scottsdale, AZ. https:\/\/arxiv.org\/abs\/1301.3781 <\/span><\/p>\n

Miller, G. A., Beckwith, R., Fellbaum, C., Gross, D., & Miller, K. J. (1990). Introduction to WordNet: An on-line lexical database. International Journal of Lexicography, 3<\/i>(4), 235\u2013244. <\/span><\/p>\n

Moon, S., Potdar, S., & Martin, L. (2014). Identifying student leaders from MOOC discussion forums through language influence. EMNLP 2014<\/i>, 15. http:\/\/www.aclweb.org\/anthology\/W14-4103 <\/span><\/p>\n

Pennebaker, J. W., Booth, R. J., & Francis, M. E. (2007). Linguistic inquiry and word count: LIWC [Computer software]. Austin, TX: liwc.net. <\/span><\/p>\n

Pennebaker, J. W., Boyd, R. L., Jordan, K., & Blackburn, K. (2015). The development and psychometric properties of LIWC2015. UT Faculty\/Researcher Works. https:\/\/repositories.lib.utexas.edu\/bitstream\/handle\/2152\/31333\/LIWC2015_LanguageManual.pdf?sequence=3<\/span><\/p>\n

Roscoe, R. D., & McNamara, D. S. (2013). Writing Pal: Feasibility of an intelligent writing strategy tutor in the high school classroom. Journal of Educational Psychology, 105<\/i>, 1010\u20131025. <\/span><\/p>\n

Rudner, L. M., Garcia, V., & Welch, C. (2006). An evaluation of IntelliMetric’ essay scoring system. The Journal of Technology, Learning and Assessment, 4<\/i>(4). https:\/\/ejournals.bc.edu\/ojs\/index.php\/jtla\/article\/download\/1651\/1493 <\/span><\/p>\n

Seaton, D. T., Bergner, Y., Chuang, I., Mitros, P., & Pritchard, D. E. (2014). Who does what in a massive open online course? Communications of the ACM, 57<\/i>(4), 58\u201365. <\/span><\/p>\n

Shermis, M. D., Burstein, J., Higgins, D., & Zechner, K. (2010). Automated essay scoring: Writing assessment and instruction. International Encyclopedia of Education, 4<\/i>, 20\u201326. <\/span><\/p>\n

Stahl, G. (2006). Group cognition: Computer support for building collaborative knowledge (pp. 451\u2013473). Cambridge, MA: MIT Press. <\/span><\/p>\n

Teplovs, C. (2008). The knowledge space visualizer: A tool for visualizing online discourse. In G. Kanselaar, V. Jonker, P. A. Kirschner, & F. Prins (Eds.), Proceedings of the International Society of the Learning Sciences 2008: Cre8 a learning world. <\/i>Utrecht, NL: International Society of the Learning Sciences. http:\/\/chris.ikit.org\/ksv2.pdf <\/span><\/p>\n

Trausan-Matu, S., Rebedea, T., Dragan, A., & Alexandru, C. (2007). Visualisation of learners’ contributions in chat conversations. In J. Fong & F. L. Wang (Eds.), Blended learning <\/i>(pp. 217\u2013226). Singapur: Pearson\/Prentice Hall. <\/span><\/p>\n

Trausan-Matu, S., Stahl, G., & Sarmiento, J. (2007). Supporting polyphonic collaborative learning. Indiana University Press, E-service Journal, 6<\/i>(1), 58\u201374. <\/span><\/p>\n

Turing, A. M. (1950). Computing machinery and intelligence. Mind, 59, 433\u2013460. http:\/\/www.loebner.net\/ Prizef\/TuringArticle.html <\/span><\/p>\n

Valenti, S., Neri, F., & Cucchiarelli, A. (2003). An overview of current research on automated essay grading. Journal of Information Technology Education: Research, 2<\/i>(1), 319\u2013330. <\/span><\/p>\n

Varner, L. K., Roscoe, R. D., & McNamara, D. S. (2013). Evaluative misalignment of 10th-grade student and teacher criteria for essay quality: An automated textual analysis. Journal of Writing Research, 5<\/i>, 35\u201359. <\/span><\/p>\n

Weigle, S.C. (2013). English as a second language writing and automated essay evaluation. In M. D. Shermis & J. Burstein (Eds.), Handbook of automated essay evaluation: Current applications and new directions <\/i>(pp. 36\u201354). London: Routledge. <\/span><\/p>\n

Wen, M., Yang, D., & Ros\u00e9, C. P. (2014a). Linguistic reflections of student engagement in massive open online courses. In E. Adar & P. Resnick (Eds.), Proceedings of the 8th International AAAI Conference on Weblogs and Social Media <\/i>(ICWSM\u201914), 1\u20134 June 2014, Ann Arbor, Michigan, USA. Palo Alto, CA: AAAI Press. http:\/\/www. aaai.org\/ocs\/index.php\/ICWSM\/ICWSM14\/paper\/viewFile\/8057\/8153 <\/span><\/p>\n

Wen, M., Yang, D., & Ros\u00e9, C. P. (2014b). Sentiment analysis in MOOC discussion forums: What does it tell us? In J. Stamper, Z. Pardos, M. Mavrikis, & B. M. McLaren (Eds.), Proceedings of the 7th International Conference on Educational Data Mining <\/i>(EDM2014), 4\u20137 July, London, UK (pp. 185\u2013192). International Educational Data Mining Society. <\/span><\/p>\n

Wilson, M.D. (1988). The MRC psycholinguistic database: Machine-readable dictionary (Version 2). Behavioral Research Methods, Instruments, and Computers, 201<\/i>, 6\u201311. <\/span><\/p>\n

Wong, B. Y. L. (1985). Self-questioning instructional research: A review. Review of Educational Research, 55<\/i>, 227\u2013268. <\/span><\/p>\n

Xi, X. (2010). Automated scoring and feedback systems: Where are we and where are we heading? Language Testing, 27<\/i>(3), 291\u2013300.<\/p>\n

\n
\n

1<\/a> orj. descriptive statistics<\/span><\/p>\n<\/div>\n

\n

2<\/a> orj. discriminant<\/span><\/p>\n<\/div>\n","protected":false},"author":1,"menu_order":4,"template":"","meta":{"pb_show_title":"on","pb_short_title":"","pb_subtitle":"","pb_authors":[],"pb_section_license":""},"chapter-type":[48],"contributor":[],"license":[],"class_list":["post-52","chapter","type-chapter","status-publish","hentry","chapter-type-numberless"],"part":46,"_links":{"self":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters\/52","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters"}],"about":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/types\/chapter"}],"author":[{"embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/users\/1"}],"version-history":[{"count":0,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters\/52\/revisions"}],"part":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/parts\/46"}],"metadata":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters\/52\/metadata\/"}],"wp:attachment":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/media?parent=52"}],"wp:term":[{"taxonomy":"chapter-type","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapter-type?post=52"},{"taxonomy":"contributor","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/contributor?post=52"},{"taxonomy":"license","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/license?post=52"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}