{"id":126,"date":"2020-09-03T16:40:05","date_gmt":"2020-09-03T13:40:05","guid":{"rendered":"http:\/\/acikkitap.com.tr\/oaek\/chapter\/bolum-27-ogrenme-analitigi-ve-egitsel-veri-madenciligi-uzerine-elestirel-bir-bakis\/"},"modified":"2020-09-03T16:40:05","modified_gmt":"2020-09-03T13:40:05","slug":"bolum-27-ogrenme-analitigi-ve-egitsel-veri-madenciligi-uzerine-elestirel-bir-bakis","status":"publish","type":"chapter","link":"https:\/\/acikkitap.com.tr\/oaek\/chapter\/bolum-27-ogrenme-analitigi-ve-egitsel-veri-madenciligi-uzerine-elestirel-bir-bakis\/","title":{"raw":"B\u00f6l\u00fcm 27 \u00d6\u011frenme Analiti\u011fi ve E\u011fitsel Veri Madencili\u011fi \u00dczerine Ele\u015ftirel Bir Bak\u0131\u015f","rendered":"B\u00f6l\u00fcm 27 \u00d6\u011frenme Analiti\u011fi ve E\u011fitsel Veri Madencili\u011fi \u00dczerine Ele\u015ftirel Bir Bak\u0131\u015f"},"content":{"raw":"\n

Rita Kop1<\/sup>, Helene Fournier2<\/sup>, Guillaume Durand2 <\/sup><\/span><\/span><\/p>\n

1<\/sup>E\u011fitim Fak\u00fcltesi, Yorkville \u00dcniversitesi, Birle\u015fik Krall\u0131k<\/span><\/span><\/p>\n

2<\/sup>Bilgi ve \u0130leti\u015fim Teknolojileri, Kanada Ulusal Ara\u015ft\u0131rma Konseyi, Kanada<\/span><\/span><\/p>\n

DOI: 10.18608\/hla17.027<\/span><\/span><\/p>\n\n

\u00d6Z<\/h2>\n

E\u011fitsel veri madencili\u011fi (EVM) ve \u00f6\u011frenme analiti\u011fi (\u00d6A; Fournier, Kop ve Durand, 2014) hakk\u0131ndaki son yaz\u0131m\u0131zda, nicel ve nitel analiz ara\u00e7lar\u0131n\u0131n kullan\u0131labilirli\u011fi ile ilgili yay\u0131nlar\u0131n hen\u00fcz mevcut olmad\u0131\u011f\u0131 ve \u00f6\u011frenenlere \u00f6z-y\u00f6netimli \u00f6\u011frenme yolculuklar\u0131nda yard\u0131mc\u0131 olabilecek daha fazla ara\u015ft\u0131rman\u0131n yararl\u0131 olaca\u011f\u0131 sonucuna varm\u0131\u015ft\u0131k. Bahsi ge\u00e7en ara\u015ft\u0131rmalardan baz\u0131lar\u0131n\u0131n tekrar edilmesi hayal k\u0131r\u0131kl\u0131\u011f\u0131na yol a\u00e7an sonu\u00e7lar verse de bu yay\u0131nlardan baz\u0131lar\u0131 ger\u00e7ekle\u015ftirilmi\u015ftir. Bu b\u00f6l\u00fcmde, e\u011fitim ve \u00f6\u011frenim ortamlar\u0131n\u0131n sonu\u00e7lar\u0131n\u0131 \u00f6l\u00e7mek ve talep etmek i\u00e7in EVM'nin ve \u00d6A'n\u0131n ge\u00e7erlili\u011fi konusunda ele\u015ftirel bir tutum sergiliyoruz. Ayr\u0131ca, deneysel \u00f6\u011frenme modellerinin yanl\u0131\u015fl\u0131klar\u0131n\u0131 g\u00f6stermek i\u00e7in EVM'nin nas\u0131l kullan\u0131labilece\u011fini de rapor edece\u011fiz. Ara\u015ft\u0131r\u0131lacak di\u011fer boyutlar; \u00f6\u011frenmede insan fakt\u00f6rleri, bu fakt\u00f6rlerin EVM ve \u00d6A ile ili\u015fkileri ve a\u00e7\u0131k \u00f6\u011frenme ortamlar\u0131ndaki ara\u015ft\u0131rmalarda \u201cB\u00fcy\u00fck Veri\u201dyi kullanma eti\u011fidir.<\/span><\/span><\/p>\n

Anahtar Kelimeler<\/span>: E\u011fitsel veri madencili\u011fi (EVM), b\u00fcy\u00fck veri, algoritmalar, rastlant\u0131<\/span><\/span><\/p>\n

Son on y\u0131l i\u00e7inde e\u011fitim ve \u00f6\u011fretim teknolojisi alan\u0131 s\u0131ra d\u0131\u015f\u0131 bir de\u011fi\u015fim ve geli\u015fim g\u00f6stermektedir. E\u011fitim alan\u0131nda yap\u0131lan ge\u00e7mi\u015f ara\u015ft\u0131rmalar; \u00f6\u011frenen, \u00f6\u011freten ve kurs i\u00e7eriklerinden olu\u015fan bir e\u011fitim \u00fc\u00e7geniyle ili\u015fkiliyken (Kansanen ve Meri, 1999; Meyer ve Land, 2006) yeni geli\u015ftirilen teknolojiler \u00f6\u011frenmeyi etkileyen di\u011fer boyutlara vurgu yapmaktad\u0131r. \u00d6rne\u011fin, \u00f6\u011frenme ba\u011flam\u0131 veya \u00f6\u011frenme ortam\u0131 ve kullan\u0131lmakta olan teknolojiler (Bouchard, 2013). Fenwick (2015a), insanlar\u0131n ve kulland\u0131klar\u0131 teknolojilerin birbirinden ba\u011f\u0131ms\u0131z de\u011ferlendirilemeyece\u011fini \u015fu \u015fekilde iddia etmektedir: \u201cMateryal ve sosyal g\u00fc\u00e7ler, e\u011fitim s\u00fcre\u00e7lerinin ve etkinliklerinin ortak olu\u015fumunu nas\u0131l de\u011ferlendirmemiz gerekti\u011fi hususunda \u00f6nemli uygulamalar\u0131 kapsayacak \u015fekilde i\u00e7 i\u00e7edirler\u201d (s. 14). \u0130nsanlar ve teknoloji gibi materyaller aras\u0131nda sadece bir etkile\u015fim de\u011fil, ayn\u0131 zamanda birlikte ya\u015fayan bir ili\u015fki vard\u0131r.<\/span><\/p>\n

Yeni teknolojiler bizi bilgi k\u0131tl\u0131\u011f\u0131 \u00e7a\u011f\u0131ndan, bilgi bollu\u011fu \u00e7a\u011f\u0131na ta\u015f\u0131m\u0131\u015ft\u0131r (Weller, 2011). Art\u0131k sosyal medya, a\u011flar \u00fczerinden k\u00fcresel \u00f6l\u00e7ekte, tu\u011fla duvarlarla s\u0131n\u0131rlanan geleneksel s\u0131n\u0131f d\u0131\u015f\u0131nda ileti\u015fim kurmay\u0131 m\u00fcmk\u00fcn k\u0131l\u0131yor. Bu kadar k\u00fcresel bir \u00f6l\u00e7ekte ileti\u015fim, \u00e7ok uzun zaman \u00f6nce d\u00fc\u015f\u00fcn\u00fclemezdi. Veri ve veri depolama, geli\u015fen teknolojilerin etkisi alt\u0131nda geli\u015fmi\u015ftir. Veri toplamak ve veritaban\u0131na kaydetmek yerine, \u015fimdi algoritmalar ve makine \u00f6\u011frenimi kullan\u0131larak temsil edilip g\u00f6rselle\u015ftirilebilen bulutta depolanan b\u00fcy\u00fck veri ak\u0131\u015flar\u0131yla ilgileniyoruz. Bu veriden \u00f6\u011frenmek i\u00e7in ilgin\u00e7 f\u0131rsatlar sunar, bununla birlikte gizli kavray\u0131\u015flar\u0131 ve ayn\u0131 zamanda \u00f6nemli zorluklar\u0131 ortaya \u00e7\u0131kar\u0131r.<\/span><\/p>\n

E\u011fitim s\u00fcrecindeki payda\u015flar\u0131n -\u00f6\u011frenciler, e\u011fitimciler ve y\u00f6neticiler- t\u00fcm bu bilgi seviyelerini etkili bir \u015fekilde eri\u015febilmelerini, y\u00f6netebilmelerini ve anlamlar\u0131n\u0131 sa\u011flayabilecekleri ile ilgili sorular sorulmu\u015ftur. Bilgisayar bilimciler, tam olarak bunu yapabilen otomatik veri filtreleme ve analizi i\u00e7in f\u0131rsatlar \u00f6nermi\u015ftir: mevcut t\u00fcm verileri elden ge\u00e7irin ve \u00f6\u011frenenlere ba\u011flant\u0131lar sa\u011flay\u0131n ve tercih ettikleri bilgiler, insanlar ve ara\u00e7lar i\u00e7in tavsiyeler ve bunu yaparken \u00f6\u011frenim deneyimini ki\u015fiselle\u015ftirin ve \u00f6\u011frenenlerin \u00f6\u011frenmelerinin y\u00f6netimi ve derinle\u015ftirilmesinde \u00f6\u011frenenlere yard\u0131mc\u0131 olun (Siemens, Dawson ve Lynch, 2013). Ayr\u0131ca, \u00f6\u011frenenlerin etkinliklerinin geride b\u0131rakt\u0131\u011f\u0131 izlerden elde edilen verilere eri\u015filerek b\u00fcy\u00fck kurumsal veri k\u00fcmeleri kullanan ara\u015ft\u0131rma \u00f6rnekleri de ortaya \u00e7\u0131kmaktad\u0131r. (Xu ve Smith Jaggars, 2013).<\/span><\/p>\n

De\u011fi\u015fimlerin tart\u0131\u015f\u0131lmas\u0131nda b\u00fcy\u00fck veriler profesyonel uygulamaya zorlayabilir, \u00f6rne\u011fin Fenwick (2015b), karar verme a\u00e7\u0131s\u0131ndan \"bilginin azalt\u0131lmas\u0131n\u0131 vurgulamaktad\u0131r. Veri analiti\u011fi yaz\u0131l\u0131m\u0131 sorunlar\u0131n teknik oldu\u011fu, bilinebilir, \u00f6l\u00e7\u00fclebilir parametrelerden olu\u015ftu\u011funu ve teknik hesaplama yoluyla \u00e7\u00f6z\u00fclebilece\u011fi basit \u00f6nc\u00fclleri ile \u00e7al\u0131\u015f\u0131r. Etik ve de\u011ferlerin karma\u015f\u0131kl\u0131\u011f\u0131, belirsizliklerin ve gerilimlerin, k\u00fclt\u00fcr\u00fcn ve politikalar\u0131n ve hatta verilerin topland\u0131\u011f\u0131 ba\u011flam dikkate al\u0131nmaz \u201d(s. 70). <\/span><\/p>\n

Bu \u00f6nemli bir konudur. Ayr\u0131ca, verileri i\u00e7eren mevcut geli\u015fmeler uyguland\u0131\u011f\u0131nda, g\u00fcnl\u00fck prati\u011fimizi tam olarak anla\u015f\u0131lamayacak \u015fekillerde de\u011fi\u015ftirebilece\u011fini vurgulamaktad\u0131r. \u00d6rne\u011fin, kar\u015f\u0131la\u015ft\u0131rma ve tahminde bir ba\u011f\u0131ml\u0131l\u0131k oldu\u011funda ortaya \u00e7\u0131kan e\u015fitlik sorunlar\u0131na dikkat \u00e7ekmektedir (Fenwick, 2015b). Ayr\u0131ca, onun ara\u015ft\u0131rmas\u0131, potansiyel e\u011fitimcilere ve e\u011fitim ara\u015ft\u0131rmac\u0131lar\u0131na \u00f6\u011fretilen ara\u015ft\u0131rma metodolojilerinin, uygulamalar\u0131n\u0131 geli\u015ftirmek i\u00e7in mevcut b\u00fcy\u00fck veri k\u00fcmeleriyle ba\u015fa \u00e7\u0131kmada tamamen yetersiz oldu\u011fu sonucuna varm\u0131\u015ft\u0131r. Ayr\u0131ca, \u201cprofesyonel eylemlili\u011fin ve hesap verilebilirli\u011fin seviyesini\" merak etmektedir. \u00c7ok fazla veri birikimi ve hesaplamas\u0131 otomatikle\u015ftirilmi\u015ftir, bu da algoritmalar\u0131n \u00f6zerkli\u011fi ve k\u00f6t\u00fc \u015feyler oldu\u011funda sorumlulu\u011fu \u00fcstlenmek hakk\u0131nda yeni sorular ortaya \u00e7\u0131karmaktad\u0131r \u201d(s. 71). Bunlar dikkatle d\u00fc\u015f\u00fcn\u00fclmesi gereken ciddi sorulard\u0131r. Bu b\u00f6l\u00fcm, ara\u015ft\u0131rmalarda b\u00fcy\u00fck veri k\u00fcmelerini ve \u00f6\u011frenene destek i\u00e7in kullan\u0131c\u0131 verilerini kullanan e\u011fitsel veri madencili\u011fi ve analiti\u011fi ile ilgili baz\u0131 zorluklar\u0131 ele alacakt\u0131r. Ayn\u0131 zamanda otomasyonun etkisini ve teknolojiyle \u00f6\u011frenmede insan ileti\u015fiminin ve kat\u0131l\u0131m\u0131n\u0131n de\u011fi\u015ftirilmesiyle makinele\u015fmenin olas\u0131 etkilerini de ara\u015ft\u0131racakt\u0131r.<\/span><\/p>\n\n

\u00d6\u011eRENME Y\u00d6NET\u0130M\u0130NDE E\u011e\u0130TSEL VER\u0130 MADENC\u0130L\u0130\u011e\u0130 FIRSATLARI<\/h2>\n

G\u00fcvenilirlik ve Ge\u00e7erlilik<\/h3>\n

E\u011fitsel veri madencili\u011fi (EVM), e\u011fitim ortamlar\u0131ndan gelen benzersiz veri t\u00fcrlerini ara\u015ft\u0131rmak i\u00e7in y\u00f6ntemler geli\u015ftirmekle ve \u00f6\u011frencilerin ve \u00f6\u011frendikleri ortamlar\u0131 daha iyi anlamak i\u00e7in bu y\u00f6ntemleri kullanmakla ilgilenen bir disiplindir (Ed Tech Review, 2016). E\u011fitsel veri madencili\u011fi, ad\u0131ndan da daha geni\u015f bir kapsam\u0131 ifade etmekte olup bilgi getirimi ve \u00f6\u011frenme mekanizmalar\u0131n\u0131n daha iyi anla\u015f\u0131lmas\u0131 i\u00e7in e\u011fitim verilerinin ara\u015ft\u0131r\u0131lmas\u0131ndan da \u00f6te bir anlam\u0131 ifade etmektedir. Bu nedenle, EVM ayn\u0131 zamanda ge\u00e7erlilik, tekrar \u00fcretilebilirlik ve genelle\u015ftirilebilirlik ile ilgili bilimsel kayg\u0131larla y\u00f6netilen istatistiksel yakla\u015f\u0131mlar\u0131 kullanarak makine \u00f6\u011frenmesini ve istatistiksel yakla\u015f\u0131mlar\u0131 kullanarak \u00f6\u011frenen davran\u0131\u015flar\u0131n\u0131 tahmin etmek i\u00e7in y\u00f6ntemler ve modeller geli\u015ftirmeyi ama\u00e7lamaktad\u0131r.<\/span><\/p>\n

\u00d6\u011frenme analiti\u011fi (\u00d6A), EVM alan\u0131yla yak\u0131ndan ilgilidir ve \u00f6\u011frenmeyi ve ger\u00e7ekle\u015fti\u011fi ortamlar\u0131 anlamak ve optimize etmek amac\u0131yla \u00f6\u011frenenler ve ba\u011flamlar\u0131 hakk\u0131ndaki verilerin \u00f6l\u00e7\u00fcm\u00fc, toplanmas\u0131, analizi ve raporlanmas\u0131 ile ilgilidir (Long ve Siemens, 2011). \u00d6\u011frenme s\u00fcrecini artt\u0131rmak i\u00e7in EVM teknikleri ve \u00d6A kullan\u0131l\u0131r. Bunlar etkili \u00f6\u011frenci deste\u011fi sa\u011flanmas\u0131na yard\u0131mc\u0131 olma konusunda \u00fcmit verici g\u00f6r\u00fcnmektedir ve bu yeni geli\u015fmelerin e\u011fitimi ve \u00f6\u011frenmeyi art\u0131raca\u011f\u0131 vaadine ra\u011fmen, \u00f6nemli zorluklar da tespit edilmi\u015ftir.<\/span><\/p>\n

Bir \u00f6l\u00e7\u00fcde, EVM sadece d\u00fcnyan\u0131n d\u00f6rt bir yan\u0131ndan ara\u015ft\u0131rmac\u0131lar\u0131n \u00fcretken katk\u0131lar\u0131ndan dolay\u0131 geli\u015fen bir ara\u015ft\u0131rma alan\u0131 de\u011fil ayn\u0131 zamanda bir bilimdir. Son zamanlarda, \u0130ngiltere Bilim Konseyi bilimi \u201ckan\u0131tlara dayal\u0131 sistematik bir metodoloji izleyerek do\u011fal ve sosyal d\u00fcnyan\u0131n bilgi ve anlay\u0131\u015f\u0131 pe\u015finde olmak\u201d olarak tan\u0131mlam\u0131\u015ft\u0131r (British Science Council, 2009). Kan\u0131t, di\u011fer herhangi bir bilimsel alanda oldu\u011fu gibi alanda yap\u0131lan herhangi bir iddia i\u00e7in bir gerekliliktir; e\u011fitsel veri madencili\u011fi ve analiti\u011fi ara\u015ft\u0131rmac\u0131lar\u0131, e\u011fitim verilerinden al\u0131nan veya do\u011frulanan iddialar\u0131 ve ke\u015fifleri desteklemek veya reddetmek i\u00e7in kan\u0131ta ihtiya\u00e7 duyar.<\/span><\/p>\n

Bununla birlikte, sa\u011flam ya da zay\u0131f kan\u0131t\u0131n ne oldu\u011funa dair ortak bir tan\u0131mlama, \u201cpozitif\u201d (Bilgisayar Bilimi) ve \u201csosyal\u201d bilimlerden (E\u011fitim) bir\u00e7ok bilim insan\u0131n\u0131 bir araya getiren EVM ve \u00d6A ara\u015ft\u0131rma toplulu\u011fu i\u00e7in a\u00e7\u0131k de\u011fildir. Burada, kendi ara\u015ft\u0131rmam\u0131z s\u0131ras\u0131nda rastlad\u0131\u011f\u0131m\u0131z baz\u0131 tutars\u0131zl\u0131klar ve y\u00f6ntemsel kusurlardan \u00f6rnekler sunaca\u011f\u0131z. Veri payla\u015f\u0131m\u0131 sayesinde, Long ve Aleven (2014) bir ak\u0131ll\u0131 \u00f6\u011fretici sisteminde oyunla\u015ft\u0131r\u0131lm\u0131\u015f bir yakla\u015f\u0131m\u0131n \u00f6\u011frenme iddialar\u0131yla ters d\u00fc\u015febildiler. Ancak bazen veri k\u00fcmelerini payla\u015fmak yeterli de\u011fildir; baz\u0131 ara\u015ft\u0131rma \u00e7al\u0131\u015fmalar\u0131, bir ara\u015ft\u0131rma belgesinde a\u00e7\u0131k\u00e7a tan\u0131mlanmas\u0131 zor olabilecek ad\u0131mlar esnas\u0131nda birden \u00e7ok se\u00e7im (yanl\u0131l\u0131k) yap\u0131ld\u0131\u011f\u0131 i\u00e7in kapsaml\u0131 bir \u00f6n i\u015flemeyi gerektirir. Bu y\u00fczden ayn\u0131 kurallar\u0131 izleyerek veri k\u00fcmesini haz\u0131rlamaya \u00e7al\u0131\u015fan ba\u015fka bir ekip bunu yapmay\u0131 ba\u015faramayabilir. \u00d6n i\u015fleme s\u0131ras\u0131nda kullan\u0131lan yaz\u0131l\u0131m\u0131n niteli\u011fi de etkili olabilir. \u00d6nemli y\u00f6ntemlerin uygulanmas\u0131, R, SPSS, Matlab ve di\u011fer ara\u00e7lar\u0131 kullan\u0131rken de\u011fi\u015fiklik g\u00f6stererek potansiyel olarak farkl\u0131 \u00e7\u0131kar\u0131mlara yol a\u00e7abilir.<\/span><\/p>\n

Bir ba\u015fka ihtilafl\u0131 y\u00f6n, istatistiksel bir modelin \u00fczerine kurulaca\u011f\u0131 deney \u00f6ncesi varsay\u0131mlar<\/i><\/span> veya \u201ctemel referans de\u011fer\u201d dir. \u00d6rne\u011fin, insan uzmanlar\u0131n maddeler ve beceri e\u015flemelerini tan\u0131mlad\u0131\u011f\u0131 yeterlik \u00e7er\u00e7eveleri sorgulanabilirdir. (Durand, Belacel ve Goutte, 2015). Di\u011fer bir\u00e7ok gizli \u00f6zellik gibi becerileri nitelendirmek bazen zordur. Bu ama\u00e7la, PSLC Veri Ma\u011fazas\u0131 \u00f6\u011frenme uzmanlar\u0131na, \u00f6\u011frencilerden edinilen g\u00f6zlem sonu\u00e7lar\u0131n\u0131n yapt\u0131klar\u0131 e\u015flemeleri geli\u015ftirmeye ve payla\u015fmaya yard\u0131mc\u0131 oldu\u011fundan yeterlilik \u00e7er\u00e7evelerini test etmeleri i\u00e7in inan\u0131lmaz bir ortam sunar. Ayr\u0131ca, veri k\u00fcmelerini payla\u015fmak sorunlar\u0131 tan\u0131mlama konusunda de\u011ferli oldu\u011fundan, veri k\u00fcmelerini payla\u015fmak EVM ve \u00d6A pratisyenleri aras\u0131nda harika bir ara\u00e7t\u0131r. Payla\u015f\u0131m ola\u011fan olmal\u0131 ve yay\u0131nlanan hi\u00e7bir sonu\u00e7 di\u011fer ekiplerin bu iddialar\u0131 do\u011frulama imk\u00e2n\u0131 olmadan ciddi \u015fekilde d\u00fc\u015f\u00fcn\u00fclmemelidir.<\/span><\/p>\n

Di\u011fer baz\u0131 konular, istatistiksel \u00e7al\u0131\u015fmalar\u0131 ve \u00f6zellikle do\u011frusal korelasyon \u00f6l\u00e7\u00fclerini g\u00f6z \u00f6n\u00fcnde bulundurmak gibi sorular do\u011furabilir. EVM, neyin kan\u0131t olarak kullan\u0131labilece\u011fi veya kullan\u0131lamayaca\u011f\u0131 konusunda farkl\u0131 uygulamalar\u0131 olan ve farkl\u0131 bak\u0131\u015f a\u00e7\u0131lar\u0131na sahip ara\u015ft\u0131rmac\u0131lar\u0131 bir araya getirir. \u201cPozitif bilimlerde\u201d, r =.5'in alt\u0131ndaki \u00f6nemli bir Pearson korelasyonunun sistematik olarak zay\u0131f oldu\u011fu d\u00fc\u015f\u00fcn\u00fcl\u00fcrken, \u201csosyal bilimlerde\u201d, r =.3 de\u011ferlerinin g\u00fc\u00e7l\u00fc oldu\u011funu d\u00fc\u015f\u00fcnmek normaldir. Psikologlar bile buna.3 e\u015fi\u011fi, \u201cki\u015filik katsay\u0131s\u0131\u201d diyorlar \u00e7\u00fcnk\u00fc ki\u015filik \u00f6zellikleri ile davran\u0131\u015flar aras\u0131ndaki ili\u015fkilerin \u00e7o\u011fu, yetkinlik ve performans aras\u0131ndaki ili\u015fki de d\u00e2hil olmak \u00fczere, bu de\u011fer etraf\u0131nda olma e\u011filimindedir (Mischel, 1968, s. 78). EVM'de duygu analizi ile ilgili yap\u0131lan \u00e7al\u0131\u015fmalar (Wen, Yang ve Rose, 2014), KA\u00c7D'lerde okulu b\u0131rakma oranlar\u0131 konusunda anlaml\u0131 olan \u201csosyal\u201d bilim ara\u015ft\u0131rma sonu\u00e7lar\u0131ndan bilgi i\u015flemsel sonu\u00e7lar elde etmenin zor oldu\u011fu bir \u00f6rnek sunmaktad\u0131r. Ayr\u0131ca, incelenmekte olan konunun nitel tekniklerle daha iyi ara\u015ft\u0131r\u0131lmas\u0131 \u00f6nerilebilir. Bununla birlikte, ama\u00e7 \u00e7\u0131kar\u0131mlarda bulunmak oldu\u011funda nicel EVM formundaki ili\u015fkiler zay\u0131f kalmaktad\u0131r. \u00d6zellikle, tan\u0131m gere\u011fi kriterdeki varyans\u0131n %9'unu a\u00e7\u0131klayan bir a., 3. korelasyonu, duygu analizi alan\u0131ndaki tahminlerde s\u0131n\u0131rl\u0131 bir de\u011fere sahip olabilir.<\/span><\/p>\n

Birka\u00e7 istatistik testi de anlaml\u0131 olabilir ancak sonu\u00e7lar\u0131n kesinli\u011fi konusunda tam olarak ger\u00e7e\u011fe uygun olmayabilir. El Emam (1998), bir Ki-Kare testinin bir s\u0131n\u0131fland\u0131r\u0131c\u0131n\u0131n \u00f6ng\u00f6r\u00fcc\u00fc ge\u00e7erlili\u011fini de\u011ferlendirmede nas\u0131l yan\u0131lt\u0131c\u0131 olabilece\u011fini de\u011ferlendirmi\u015ftir. Daha yak\u0131n zamanlarda, Gonzalez-Brenes ve Huang (2015), Leopard \u00f6l\u00e7\u00fcm\u00fcn\u00fc uyarlanabilir \u00f6zel ders sistemlerini de\u011ferlendirmenin ve onlar\u0131n faydal\u0131l\u0131klar\u0131n\u0131 de\u011ferlendirerek sistemin \u00f6ng\u00f6r\u00fcc\u00fc do\u011frulu\u011funu de\u011ferlendirme sonu\u00e7lar\u0131n\u0131 art\u0131rman\u0131n standart bir yolu olarak \u00f6nermi\u015ftir. Onlar bu sistemlerdeki \u00f6\u011frenme \u00e7\u0131kt\u0131lar\u0131na eri\u015fmek i\u00e7in \u00f6\u011frenenlerden istenecek \u00e7aba miktar\u0131n\u0131 de\u011ferlendirmeyi teklif ettiler. Sonu\u00e7ta, fayda \u00f6l\u00e7\u00fcleri, sistemleri kullanan ki\u015filerin en \u00e7ok ilgilendikleri \u015fey olabilir.<\/span><\/p>\n

Bu ama\u00e7la, EVM toplulu\u011funun \u00f6nde gelen ara\u015ft\u0131rmac\u0131lar\u0131ndan Ryan Baker, KA\u00c7D\u2019de \u201cE\u011fitimde B\u00fcy\u00fck Veri\u201d ba\u015fl\u0131kl\u0131 ara\u015ft\u0131rmas\u0131nda, ara\u015ft\u0131rmac\u0131lar\u0131n modellerinin ge\u00e7erlili\u011fini anlamalar\u0131na ve daha ak\u0131lc\u0131 bir \u015fekilde kontrol etmelerine yard\u0131mc\u0131 olacak \u00f6rnekler ve iyi uygulama \u00f6nerileri sunmaktad\u0131r (genelle\u015ftirilebilirlik, ekolojik, yap\u0131 ve tahmine dayal\u0131, esas ve i\u00e7erik ge\u00e7erlili\u011fi). Baker bu dersinde, kesinlik, ROC (e\u011frilik yar\u0131\u00e7ap\u0131), keskinlik ve hassasiyet veya Ki-Kare gibi s\u0131n\u0131fland\u0131r\u0131c\u0131lara ili\u015fkin di\u011fer \u00f6l\u00e7\u00fclerin kusurlar\u0131n\u0131 bertaraf etmede nas\u0131l \u201cduyargan\u0131n \u015fanstan daha iyi oldu\u011funu\u201d ve belli bir \u00f6zelli\u011fi \u201cduyargan\u0131n do\u011fru bir \u015fekilde tespit etme olas\u0131l\u0131\u011f\u0131n\u0131\u201d \u00f6l\u00e7mek i\u00e7in s\u0131ras\u0131yla Kappa'y\u0131 ve hatta daha iyi olan A'y\u0131 kulland\u0131\u011f\u0131n\u0131 vurgulad\u0131. (Ocumpaugh, Baker, Gowda, Heffernan ve Heffernan, 2014, s. 492). Ancak EVM yay\u0131nlar\u0131nda A\u2019 ve Kappa kullan\u0131m\u0131 \u015fu ana kadar s\u0131n\u0131rl\u0131 g\u00f6r\u00fcnmektedir.<\/span><\/p>\n

Ara\u015ft\u0131rma b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fcn son derece \u00f6nemli oldu\u011funu vurgulamak istiyoruz. EVM ve \u00d6A'da \u201cuydurulmu\u015f\u201d sonu\u00e7lar vermek \u00e7ekici olabilir. Kendi \u00e7al\u0131\u015fmam\u0131z, somut sonu\u00e7lar\u0131n elde edilmesinin genellikle bir\u00e7ok \u00e7aba gerektirdi\u011fini, bir\u00e7ok \u00e7al\u0131\u015fman\u0131n ba\u015far\u0131 garantisi olmadan yap\u0131ld\u0131\u011f\u0131n\u0131 ve do\u011frulama s\u00fcrecinin sorunlu g\u00f6r\u00fcnd\u00fc\u011f\u00fcn\u00fc g\u00f6stermi\u015ftir. Bug\u00fcne kadar ger\u00e7e\u011fi \u00e7arp\u0131tacak \u00f6nemli durumlar ortaya \u00e7\u0131kmam\u0131\u015ft\u0131r ancak di\u011fer bilimsel alanlarda g\u00f6zlemlendi\u011fi gibi gelecekteki olas\u0131 sahtek\u00e2rl\u0131k ve suistimal davalar\u0131ndan ka\u00e7\u0131nmak i\u00e7in \u015feffafl\u0131\u011fa ili\u015fkin y\u00f6nergeler sunma konusu daha fazla dikkate al\u0131nmal\u0131d\u0131r (Gupta, 2013).<\/span><\/p>\n

\u00d6A ve EVM'nin geli\u015fmekte olan alanlar\u0131nda, bilimsel hedefin iddial\u0131 oldu\u011funu ve ara\u015ft\u0131rmac\u0131lar\u0131n iddialar\u0131n\u0131n bilimsel sa\u011flaml\u0131\u011f\u0131n\u0131 dikkatlice kontrol etmelerini gerektirdi\u011fini savunuyoruz. Ara\u015ft\u0131rmalar yak\u0131n gelecekte insan \u00f6\u011frenmesini etkileyecekleri s\u00f6z\u00fcne dair verilen ba\u011f\u0131\u015flar ile desteklense de \u00d6A ve EVM alanlar\u0131n\u0131n bilimsel d\u00fcr\u00fcstl\u00fc\u011f\u00fc korumak i\u00e7in zaman ay\u0131rmalar\u0131 \u00f6nemlidir. Bu kullan\u0131lan y\u00f6ntemlerin ve elde edilen sonu\u00e7lar\u0131n taraf\u0131m\u0131zca dikkatle de\u011ferlendirilmesini gerektirir.<\/span><\/p>\n\n

Nitel Veri Analizinin Zorluklar\u0131<\/h3>\n

E\u011fitim ara\u015ft\u0131rmalar\u0131n\u0131n son on y\u0131ldaki geli\u015fimine bakacak olursak, nicelden nitel ara\u015ft\u0131rmalara do\u011fru belirgin bir kay\u0131\u015f vard\u0131r (Gergen, Josselson ve Freeman, 2015). Psikologlar, \u00f6\u011frenme ve bilmenin karma\u015f\u0131kl\u0131\u011f\u0131n, bireylerin davran\u0131\u015flar\u0131n\u0131 tek ba\u015f\u0131na test ederek belirleyemedikleri fikrini giderek daha fazla desteklemektedir. \u00d6\u011frenenlerin i\u00e7inde ya\u015fad\u0131klar\u0131 topluma dair eylemlerinin ve d\u00fc\u015f\u00fcncelerinin zenginli\u011fi ve bilgi a\u011flar\u0131nda yer alan ki\u015filerle ileti\u015fim kurma \u00e7al\u0131\u015fmalar\u0131, insanlar\u0131n bilgi geli\u015ftirme ve \u00f6\u011frenmeleri hakk\u0131nda daha derin, daha kapsay\u0131c\u0131 ve son derece k\u00fclt\u00fcrel bir anlay\u0131\u015f sa\u011flar (Christopher, Wendt, Marecek ve Goodman, 2014; Denzin ve Lincoln, 2011; Gergen vd., 2015).<\/span><\/p>\n

Mevcut teknoloji y\u00f6n\u00fcnden zengin \u00f6\u011frenme ortam\u0131, sadece duvarl\u0131 bir s\u0131n\u0131f de\u011fildir, ayn\u0131 zamanda k\u00fcresel a\u011f ileti\u015fimini de i\u00e7erir; bunlar, ara\u015ft\u0131rmac\u0131lara ara\u015ft\u0131rma metodolojilerini yeniden icat etmeleri konusunda zorlayan, d\u00f6n\u00fc\u015fl\u00fc hik\u00e2yeleme ve zengin imgeleri kapsar. T\u00fcm bunlar\u0131n \u00f6tesine ge\u00e7mek -\u00f6\u011frencilerin memnuniyetini ortaya \u00e7\u0131karmak i\u00e7in yap\u0131lan ders sonu anketler, \u00f6\u011frenenler taraf\u0131ndan \u00fcretilen verileri ara\u015ft\u0131r\u0131lmas\u0131, \u00e7evrimi\u00e7i \u00f6\u011frenme deneyimi s\u0131ras\u0131nda \u00fcretilen anlat\u0131lar, g\u00f6r\u00fcnt\u00fcler ve g\u00f6rselle\u015ftirmeleri analiz etmek- \u00f6\u011frenme etkile\u015fimlerinin zengin dokusunu anlamak i\u00e7in se\u00e7enekler sunar. Art\u0131k \u00f6\u011frenme ortam\u0131n\u0131n bir par\u00e7as\u0131n\u0131 olu\u015fturan sosyal medya da de\u011fi\u015fen kelime ve resim gruplar\u0131ndaki temel boyutlar\u0131 analiz etmek, \u00f6\u011frenme s\u00fcrecinin kalbine resm\u00ee ders de\u011ferlendirmelerinden \u00e7ok daha fazla girebilir.<\/span><\/p>\n

\u0130ki kitlesel a\u00e7\u0131k \u00e7evrimi\u00e7i dersi (KA\u00c7D), PLENK2010 ve CLOM REL 2014 ara\u015ft\u0131rmas\u0131, bu ara\u015ft\u0131rman\u0131n i\u00e7erdi\u011fi zorluklar\u0131n alt\u0131n\u0131 \u00e7izmektedir (Fournier ve Kop, 2015; Kop, Fournier ve Durand, 2014). \u00d6nceki KA\u00c7D ara\u015ft\u0131rmas\u0131, verilerdeki kal\u0131plar\u0131 g\u00f6rselle\u015ftirmek i\u00e7in g\u00fc\u00e7l\u00fc ara\u00e7lar\u0131, \u00f6zellikle de dijital sosyal a\u011flarda her zamankinden daha b\u00fcy\u00fck ve daha zengin veri k\u00fcmeleri sa\u011flad\u0131. Bununla birlikte, bu t\u00fcr kal\u0131plar\u0131 a\u00e7\u0131\u011fa \u00e7\u0131karma \u00e7al\u0131\u015fmalar\u0131, verilerin \u00fcretildi\u011fi pedagojik ve teknik ba\u011flamdaki cevaplardan daha fazla soru sa\u011flam\u0131\u015ft\u0131r. KA\u00c7D kat\u0131l\u0131mc\u0131lar\u0131n\u0131n neden yapt\u0131klar\u0131 verileri \u00fcrettiklerini anlamaya \u00e7al\u0131\u015f\u0131rken nitel bir yakla\u015f\u0131ma do\u011fru ilerlemek, b\u00fcy\u00fck veri, EVM ve \u00d6A'n\u0131n bize karma\u015f\u0131k \u00f6\u011frenme s\u00fcre\u00e7leri ve deneyimleri hakk\u0131nda neler s\u00f6yleyebilece\u011fi ve s\u00f6yleyemedi\u011fi<\/i><\/span> \u00fczerine ele\u015ftirel bir derin d\u00fc\u015f\u00fcnmeye neden oldu.<\/span><\/p>\n

Boyd (2010) bunu \u015f\u00f6yle ifade etmi\u015ftir:<\/span><\/p>\n\n

B\u00fcy\u00fck Veri'yi \u00e7evreleyen co\u015fkunun \u00e7o\u011fu, bir parma\u011f\u0131n\u0131z\u0131 t\u0131klatarak b\u00fcy\u00fck miktarda veriye kolayca eri\u015febilmekten kaynaklan\u0131yor. Ya da Vint Cerf in ifadesiyle, \u201c\u0130nsanl\u0131k tarihinde hi\u00e7bir zaman, bu kadar \u00e7abuk ve bu kadar kolay bilgiye ula\u015famad\u0131k\u201d Ne yaz\u0131k ki, bu heyecanda kaybolan \u015fey, bu verilerin ne oldu\u011funun ve ne anlama geldi\u011finin ele\u015ftirel bir analizidir. (s. 2).<\/div>\n

Bu kadar \u00e7ok veri ve bilgi ile \u00e7ok h\u0131zl\u0131 bir \u015fekilde u\u011fra\u015f\u0131rken, ara\u015ft\u0131rmac\u0131lar\u0131n verileri, hedef kitle taraf\u0131ndan faydal\u0131 ve eri\u015filebilir bir bi\u00e7imde sindirebilecekleri anla\u015f\u0131labilir, t\u00fcketilebilir veya i\u015flem yap\u0131labilir sunum bi\u00e7imlerine d\u00f6n\u00fc\u015ft\u00fcrmek i\u00e7in en uygun s\u00fcre\u00e7leri ve teknikleri tasarlamalar\u0131 gerekir. Karma\u015f\u0131k fikirleri etkili bir \u015fekilde ula\u015ft\u0131rma yetene\u011fi, ara\u015ft\u0131rma bulgular\u0131n\u0131 prati\u011fe d\u00f6n\u00fc\u015ft\u00fcren de\u011ferli bir \u015fey \u00fcretmede kritik \u00f6neme sahiptir. E\u011fitim s\u00fcrecindeki payda\u015flar\u0131n (\u00f6r. \u00d6\u011frenciler, e\u011fitimciler ve y\u00f6neticiler) t\u00fcm bu bilgi seviyelerine nas\u0131l etkili bir \u015fekilde eri\u015febilecekleri, y\u00f6netebilecekleri ve anlam \u00e7\u0131karacaklar\u0131 ile ilgili sorular sorulmu\u015ftur; EVM ve \u00d6A y\u00f6ntemleri, tam olarak otomatik veri filtreleme ve analizin bunu nas\u0131l yapabilece\u011fine i\u015faret etmektedir. Bu \u00f6\u011frenme ve \u00f6\u011frenenler hakk\u0131nda potansiyel olarak zengin \u00e7\u0131kar\u0131mlara yol a\u00e7abilir ancak ayn\u0131 zamanda s\u00fcre\u00e7te bir\u00e7ok yeni ilgin\u00e7 ara\u015ft\u0131rma sorusu ve zorlu\u011fu da beraberinde getirebilir. Bunu yaparken, ara\u015ft\u0131rmac\u0131lar verinin ne kadar anlaml\u0131 oldu\u011funu g\u00f6stermenin yan\u0131 s\u0131ra, duyarl\u0131 ara\u015ft\u0131rma tasar\u0131mlar\u0131 ve uygulamalar\u0131 ile sorumlu inovasyon ile me\u015fgul olarak e\u011fitim s\u00fcrecinde \u00e7e\u015fitli payda\u015flara hitap etmek i\u00e7in \u00e7aba g\u00f6stermelidirler (Berland, Baker ve Blikstein, 2014).<\/span><\/p>\n\n

Algoritmalar, Mutlu Tesad\u00fcf ve \u00d6\u011frenmede \u201c\u0130nsan\u201d: \u00d6\u011frenme Analiti\u011fine Ele\u015ftirel Bir Bak\u0131\u015f<\/span><\/span><\/h4>\n

EVM ve \u00d6A'da yava\u015f yava\u015f bir alan yaz\u0131n geli\u015fmektedir. Temel olarak, \u00f6\u011frenmeyi analiz etmek i\u00e7in teknolojiyi kullanmak ya da \u00f6\u011frenmeyi ilerletmek i\u00e7in yorday\u0131c\u0131 analiti\u011fi kullanmak kolay de\u011fildir. E\u011fitimde algoritmalar\u0131n ve di\u011fer veri g\u00fcd\u00fcml\u00fc sistemlerin geli\u015ftirilmesine ili\u015fkin konular, bu sistemlerin ger\u00e7ekte neleri de\u011fi\u015ftirdi\u011fi ve bu de\u011fi\u015fimin olumlu mu olumsuz mu oldu\u011fu ile ilgili sorulara yol a\u00e7ar. \u0130kincisi, veri g\u00fcd\u00fcml\u00fc sistemlerin i\u00e7eri\u011fini kim etkiliyor ve e\u011fitim s\u00fcrecine ne gibi katk\u0131lar sa\u011flayabilir?<\/span><\/p>\n

\u00c7evrimi\u00e7i e\u011fitimde ancak ayn\u0131 zamanda ba\u011fda\u015ft\u0131r\u0131c\u0131 bir a\u011f ortam\u0131nda (Jones, Dirckinck \u2013 Holmfeld ve Lindstrom, 2006), \u00f6\u011frenme u\u011fra\u015f\u0131ndaki kat\u0131l\u0131mc\u0131lar aras\u0131ndaki ileti\u015fim ve diyalog kaliteli bir \u00f6\u011frenme deneyiminin merkezinde olmu\u015ftur. Bu insan dokunu\u015fu \u00f6\u011frenme sistemleri ve ortamlar\u0131n\u0131 geli\u015ftirmede gerekli bir bile\u015fendir (Bates, 2014). Bilgili di\u011fer ki\u015filerin mevcudiyeti ve kat\u0131l\u0131m\u0131, kat\u0131l\u0131mc\u0131lar\u0131n resm\u00ee \u00f6\u011frenme ortamlar\u0131nda ancak ayn\u0131 zamanda \u00e7evrimi\u00e7i ilgili a\u011flarda da fikirlerini, yarat\u0131c\u0131l\u0131klar\u0131n\u0131 ve d\u00fc\u015f\u00fcncelerini geni\u015fletmek i\u00e7in her zaman hayati olarak g\u00f6r\u00fclm\u00fc\u015ft\u00fcr (Jones vd., 2006).<\/span><\/p>\n

\u00d6\u011frenme i\u00e7in veri g\u00fcd\u00fcml\u00fc teknolojiler geli\u015ftirirken, bu insan unsurunu bir \u015fekilde \u00f6\u011frenme s\u00fcrecinin yarar\u0131 i\u00e7in kullanmak \u00f6nemlidir. Bu bilginin filtrelenmesinde veya Sokratik sorular\u0131n sorulmas\u0131nda, bilginin toplanmas\u0131na insan arac\u0131l\u0131\u011f\u0131yla arac\u0131l\u0131k edilmesi gerekti\u011fi anlam\u0131na gelir (Kop, 2012). Kaynaklar hakk\u0131nda bilgi ve ba\u011flant\u0131lar sa\u011flayan \u201ctakip\u00e7iler\u201din kullan\u0131c\u0131 taraf\u0131ndan se\u00e7ildi\u011fi, de\u011ferli ve g\u00fcvenilir olarak g\u00f6r\u00fcld\u00fc\u011f\u00fc Twitter gibi sosyal mikroblog sitelerinin bunu ba\u015far\u0131l\u0131 bir \u015fekilde yapt\u0131klar\u0131 g\u00f6sterilmi\u015ftir (Bista, 2014; Kop, 2012; Stewart, 2015). Algoritmalarda, bu kararlar\u0131n elde edilmesi zordur ancak belki de verilere dayanan \u00f6neri sistemleri ile, ileti\u015fimi temel alan bili\u015fsel destek ve \u00f6\u011frenme deste\u011fi uygulamalar\u0131n\u0131n bir kombinasyonu bunu kolayla\u015ft\u0131rabilir.<\/span><\/p>\n

Veri odakl\u0131 sistemlerin i\u00e7eri\u011fini kimin etkiledi\u011fini ve e\u011fitim s\u00fcrecine ne gibi bir de\u011fer katabileceklerini d\u00fc\u015f\u00fcnmek \u00f6nemlidir. Ayr\u0131ca, sadece yeni teknolojilerin sundu\u011fu f\u0131rsatlar ve verimlili\u011fi dikkate al\u0131nmakla kalmay\u0131p, ayn\u0131 zamanda, insan ileti\u015fimi ile betimlenen bir \u00f6\u011frenme ortam\u0131ndan, \u00fczerinde \u00f6\u011frenenin kontrol\u00fcn\u00fcn az oldu\u011fu veya hi\u00e7 kontrol\u00fcn\u00fcn olmad\u0131\u011f\u0131 teknik unsurlar\u0131 i\u00e7eren bir ortama ge\u00e7meye ili\u015fkin etik unsurlar da d\u00fc\u015f\u00fcn\u00fclmelidir.<\/span><\/p>\n

Algoritmalar\u0131n geli\u015ftirilmesinde \u00f6ne s\u00fcr\u00fclen sorunlardan biri, \u00f6neri veya arama sonucunun niteli\u011fini etkileyebilecek olan ara\u015ft\u0131rmac\u0131n\u0131n araca dair yanl\u0131l\u0131\u011f\u0131n\u0131n tan\u0131t\u0131lmas\u0131d\u0131r. (Hardt, 2014). Verilerle \u00e7al\u0131\u015fan ki\u015filerin do\u011fru \u015fekilde e\u011fitilmesi, her \u015feyi de\u011fi\u015ftirebilir (Fenwick, 2015b; Boyd ve Crawford, 2012). Halen uygulamalar\u0131 sosyal bilimlerde bir ge\u00e7mi\u015fe sahip olmayan bilgisayar bilimcileri ve matematik\u00e7iler \u00fcretmektedir. Boyd ve Crawford'un (2012) ikna edici bir \u015fekilde ileri s\u00fcrd\u00fc\u011f\u00fc gibi:<\/span><\/p>\n\n

Bilgi i\u015flemsel beceriler en de\u011ferli olarak konumland\u0131r\u0131ld\u0131\u011f\u0131nda, kimlerin avantajl\u0131 oldu\u011fu ve b\u00f6yle bir ba\u011flamda kimin dezavantajl\u0131 oldu\u011fu konusunda sorular ortaya \u00e7\u0131kar. Bu, bilgisayar bilimcilerinin ve sosyal bilimcilerin ikisinin de \u00f6nerecekleri de\u011ferli bak\u0131\u015f a\u00e7\u0131lar\u0131na sahip olduklar\u0131n\u0131 kabul etmek yerine \u201csay\u0131lar\u0131 okuyabilen\u201dler etraf\u0131nda kendince yeni hiyerar\u015filer olu\u015fturur. Bu \u00f6nemli \u00f6l\u00e7\u00fcde, ayn\u0131 zamanda cinsiyetlendirilmi\u015f bir b\u00f6l\u00fcnmedir. \u015eu anda bilgi i\u015flemsel beceriye sahip ara\u015ft\u0131rmac\u0131lar\u0131n \u00e7o\u011fu erkektir ve feminist tarih\u00e7iler ve bilim felsefecilerinin g\u00f6sterdi\u011fi gibi, sorular\u0131 kimlerin sordu\u011fu hangi sorular\u0131n sorulaca\u011f\u0131n\u0131 belirler (s.674).<\/div>\n

Boyd ve Crawford (2012) bilgisayar bilimcileri ve sosyal bilimcilerin \u00f6nyarg\u0131s\u0131z, y\u00fcksek nitelikli analiz ara\u00e7lar\u0131 geli\u015ftirmek i\u00e7in birlikte \u00e7al\u0131\u015fmas\u0131 gerekti\u011fini ve farkl\u0131 alanlardaki insanlarla ekip \u00e7al\u0131\u015fmas\u0131n\u0131n b\u00fcy\u00fck veri madencili\u011fi ve analizinde de verimli olabilece\u011fini \u00f6ne s\u00fcr\u00fcyor. Tabii ki, verilerin b\u00fcy\u00fcmesi ve kullan\u0131labilirli\u011fi de bunlardan faydalanmay\u0131 \u00e7ekici k\u0131lm\u0131\u015ft\u0131r ancak yine de baz\u0131 zorluklar vard\u0131r. \u00c7o\u011fu zaman insanlar bilgileri g\u00fcvendikleri kaynaklardan al\u0131rlar ancak Fenwick'in (2015b) \u00f6nerdi\u011fi gibi, yeni usullerin kullan\u0131m\u0131 \u201cg\u00fcnl\u00fck uygulama ve sorumluluklar\u0131 tam olarak tan\u0131namayacak \u015fekillerde\u201d de\u011fi\u015ftirebilir (s. 71). \u00d6rne\u011fin o, kar\u015f\u0131la\u015ft\u0131rma ve \u00f6ng\u00f6rmeye g\u00fcvenmenin b\u00fcy\u00fck veriler dikkatle kullan\u0131lmad\u0131\u011f\u0131nda \u00f6zellikle algoritmalar\u0131 \u00fcreten insanlar kal\u0131pla\u015fm\u0131\u015f kan\u0131lar\u0131n peki\u015ftirildi\u011finin fark\u0131nda de\u011fillerse, \u201ckendi kendini peki\u015ftirebilece\u011fi ve \u00fcretebilece\u011fi, s\u00fcre\u00e7 ba\u011f\u0131ml\u0131l\u0131\u011f\u0131n\u0131 artt\u0131rabilece\u011fi ve mevcut e\u015fitsizlikleri hapsedebilece\u011fini\u201d vurguluyor.<\/span><\/p>\n

Ayr\u0131ca, h\u00e2lihaz\u0131rda kullan\u0131lmakta olan algoritmalar\u0131n \u00e7o\u011funun daha derin \u00f6\u011frenme seviyelerini artt\u0131rmak veya topluma de\u011fer katmak i\u00e7in de\u011fil ekonomik kazan\u0131m i\u00e7in \u00fcretildi\u011fini hafife almamal\u0131y\u0131z. Kitchin (2015) taraf\u0131ndan iddia edildi\u011fi gibi, \u201cYaz\u0131l\u0131m yaln\u0131zca bir dizi talimat\u0131 yerine getiren bir kod sat\u0131r\u0131 de\u011fildir, bir\u00e7ok ak\u0131l\u0131n farkl\u0131 sosyal, politik ve ekonomik ili\u015fkiler i\u00e7erisine konumlanm\u0131\u015f sonucu, ko\u015fullu, ili\u015fkisel ve ba\u011flamsal olarak ortaya \u00e7\u0131kan bir sosyal \u00fcr\u00fcn\u00fc olarak anla\u015f\u0131lmal\u0131d\u0131r. \u201d(s. 5). A\u00e7\u0131k\u00e7as\u0131 otomatikle\u015ftirilmi\u015f algoritma sistemlerinin geli\u015ftirilmesinin, bir \u015feyler ters gitti\u011finde kimin sorumlu oldu\u011funu i\u015faret etmenin zor olabilece\u011fi ba\u015fka bir do\u011fal problemi bulunmaktad\u0131r.<\/span><\/p>\n\n

Baz\u0131 Etik Hususlar<\/h3>\n

A\u00e7\u0131k \u00f6\u011frenme ortamlar\u0131, g\u00fc\u00e7l\u00fc veri analizi ara\u00e7lar\u0131 ve y\u00f6ntemleri ile bir araya getirildi\u011finde, yeni ili\u015fkiler ve \u00f6\u011frenmeye destek sa\u011flaman\u0131n yan\u0131 s\u0131ra, \u00f6\u011frenenleri insan ileti\u015fimi ile nitelendirilen bir ortamdan, \u00f6\u011freneni \u00fczerinde \u00e7ok az ya da hi\u00e7 kontrol\u00fc olmad\u0131\u011f\u0131 teknik unsurlar i\u00e7eren bir ortama ta\u015f\u0131yan \u00f6nemli etik sorunlar\u0131 ve zorluklar\u0131 da vurgulamaktad\u0131r. Genel a\u011f geli\u015ftirme konusundaki ticari \u00e7aban\u0131n \u00e7o\u011fu b\u00fcy\u00fck veriden beslenmektedir ve herhangi bir yenilik\u00e7i e\u011fitim anlay\u0131\u015f\u0131ndan yoksundur (Atkinson, 2015). Biz \u201cTeknoloji \u00e7\u00f6z\u00fcmlerinin, e\u011fitime kayda de\u011fer ve s\u00fcrd\u00fcr\u00fclebilir faydalar getirmesini sa\u011flayacak anlaml\u0131 pedagojik ve yeti\u015fkin \u00f6\u011frenme teorilerine dayanan, uzmanl\u0131k ve ara\u015ft\u0131rmalar taraf\u0131ndan bilgilendirilmi\u015f \u00f6\u011frenme tasar\u0131m\u0131n\u0131n kendisi oldu\u011funu\" kabul ediyoruz (Atkinson, 2015, s. 7).<\/span><\/p>\n

EVM ve \u00d6A da d\u00e2hil olmak \u00fczere teknolojik yenili\u011fin dinamik h\u0131z\u0131 ayn\u0131 zamanda gizlili\u011fin de\u011fi\u015fime sebep olabilecek bir eyleme ge\u00e7me bi\u00e7imiyle korunmas\u0131n\u0131 gerektirir. Bu amaca ula\u015fmak i\u00e7in, EVM ve ileri analitik alan\u0131ndaki ara\u015ft\u0131rmac\u0131lar ve sistem tasar\u0131mc\u0131lar\u0131, mahremiyet g\u00fc\u00e7lendirici teknolojileri do\u011frudan \u00fcr\u00fcn ve s\u00fcre\u00e7lerine entegre eden sorumlu yenilikleri uygulamak zorundad\u0131r (Cavoukian ve Jonas, 2012). Oblinger (2012)'e g\u00f6re, \u201cAnalitik k\u00fclt\u00fcr meselesidir, bir sorgulama k\u00fclt\u00fcr\u00fcd\u00fcr: soru sormak, veri aramak, verilerin a\u00e7\u0131\u011fa vurdu\u011fu g\u00fc\u00e7l\u00fc ve zay\u0131f y\u00f6nlere kar\u015f\u0131 d\u00fcr\u00fcst olmak, bu \u00e7abalar\u0131n sonu\u00e7lar\u0131 meyvesini verirken uyum sa\u011flamakt\u0131r.\u201d(s. 98).<\/span><\/p>\n

Bunu ak\u0131lda tutarak, yeni nesil analitikleri tasarlayan ve geli\u015ftirenlerin Tasar\u0131m yoluyla Mahremiyet konusunda bilgilendirilmelerini sa\u011flaman\u0131z\u0131 \u015fiddetle tavsiye ederiz. Bu hesap verebilirlik, ara\u015ft\u0131rma do\u011frulu\u011fu, veri koruma, mahremiyet ve r\u0131zay\u0131 i\u00e7eren fark\u0131ndal\u0131k ve sorumlu uygulamay\u0131 gerektirir (Cavoukian ve Jonas, 2012; Cormack, 2015). \u00d6zel ve halka a\u00e7\u0131k veriler aras\u0131ndaki \u00e7izgi, daha fazla a\u00e7\u0131k \u00f6\u011frenme ortamlar\u0131na kat\u0131l\u0131m f\u0131rsat\u0131 yarat\u0131ld\u0131\u011f\u0131ndan ve kat\u0131l\u0131mc\u0131lar, onlar\u0131n etkinlikleri ve onlar\u0131n davran\u0131\u015flar\u0131 Facebook, Twitter, Google gibi sosyal medya ve di\u011fer \u00e7evrim i\u00e7i kullan\u0131labilir potansiyel sosyal medya ara\u00e7lar\u0131 yoluyla eri\u015filebilir oldu\u011fundan giderek bulan\u0131kla\u015fmaktad\u0131r. B\u00fcy\u00fck veri ba\u011flam\u0131nda biz, \u201c\u0130nsanlar algoritmalar\u0131n kendileri ile ilgili nas\u0131l ili\u015fkiler ve varsay\u0131mlar yaratabileceklerini ve birle\u015ftirilmi\u015f ki\u015fisel bilginin onlar\u0131n davran\u0131\u015flar\u0131 ile ilgili izinsiz ve m\u00fcdahaleci h\u00fck\u00fcmlere nas\u0131l d\u00f6nebilece\u011fini anlamak istiyor\" (s.10) \u015feklinde g\u00f6r\u00fc\u015f belirten Avrupa Veri Koruma Denet\u00e7isine (2015) kat\u0131l\u0131yoruz.<\/span><\/p>\n\n

SONU\u00c7<\/h2>\n

B\u00fcy\u00fck veri \u00e7al\u0131\u015fmalar\u0131nda do\u011fruluk, kontrol, \u015feffafl\u0131k ve g\u00fc\u00e7 ile ilgili \u00f6nemli sorular\u0131n da ele al\u0131nmas\u0131 gerekir. Pardo ve Siemens (2014), \u00e7ok uzun s\u00fcre \u00e7ok fazla veri saklaman\u0131n (\u00f6\u011frenci dijital verileri, mahremiyete duyarl\u0131 veriler d\u00e2hil) zarar verebilece\u011fini ve ki\u015fisel verileri korumak i\u00e7in emanet edilen sisteme veya kuruma g\u00fcvensizli\u011fine yol a\u00e7abilece\u011fini belirtmektedir. B\u00fcy\u00fck veri eti\u011fi konusundaki tart\u0131\u015fmalar, veri temizli\u011fi, veri se\u00e7imi ve yorumlanmas\u0131 (Boyd ve Crawford, 2012), veri analiti\u011finin istilac\u0131l\u0131k olas\u0131l\u0131\u011f\u0131 ve insan ileti\u015fiminin ve otomatik makine \u00f6\u011frenme algoritmalar\u0131 ve geri bildirimleri le ili\u015fkisinin olas\u0131 makinele\u015ftirme etkilerine ili\u015fkin metodolojik kayg\u0131lara vurgu yapm\u0131\u015ft\u0131r. Ara\u015ft\u0131rmac\u0131lar ve geli\u015ftiricilerin (Fenwick, 2015b) yararl\u0131 gelecek ad\u0131mlar\u0131 in\u015fa etmek i\u00e7in (veri madencili\u011fi ve ak\u0131ll\u0131 \u00f6\u011frenme analitikleri d\u00e2hil olmak \u00fczere) b\u00fcy\u00fck verinin sundu\u011fu imk\u00e2n ve s\u0131n\u0131rl\u0131l\u0131klar konusunda dikkatli olmalar\u0131 gerekir. Ara\u015ft\u0131rmac\u0131lar, \u00e7al\u0131\u015fmalar\u0131nda do\u011fabilecek yanl\u0131\u015f ve yanl\u0131l\u0131klar\u0131n bir k\u0131sm\u0131ndan ka\u00e7\u0131nmak ve e\u011fitim s\u00fcrecine de\u011fer katmak i\u00e7in b\u00fcy\u00fck veri ve veri g\u00fcd\u00fcml\u00fc sistemlerdeki \u00f6nemli sorunlar\u0131 ve zorluklar\u0131 ele almak i\u00e7in ekipler halinde birlikte \u00e7al\u0131\u015fmal\u0131d\u0131r.<\/span><\/p>\n\n

KAYNAK\u00c7A<\/h2>\nAtkinson, S. P. (2015). Adaptive learning and learning analytics: A new learning design paradigm. BPP Working paper. https:\/\/spatkinson.files.wordpress.com\/2015\/05\/atkinson-adaptive-learning-and-learning-analytics.pdf <\/span><\/span>\n\nBates, T. (2014). Two design models for online collaborative learning: Same or different? http:\/\/www.tonybates.ca\/2014\/11\/28\/two-design-models-for-online-collaborative-learning-same-or-different\/ <\/span><\/span>\n\nBerland, M., Baker, R. S., & Blikstein, P. (2014). Educational data mining and learning analytics: Applications to constructionist research. Technology, Knowledge and Learning, 19<\/i>, 206\u2013220. <\/span><\/span>\n\nBista, K. (2014). Is Twitter an effective pedagogical tool in higher education? Perspectives of education graduate students. Journal of the Scholarship of Teaching and Learning, 15<\/i>(2), 83\u2013102. http:\/\/files.eric.ed.gov\/ fulltext\/EJ1059422.pdf <\/span><\/span>\n\nBouchard, P. (2013). Education without a distance: Networked learning. In T. Nesbit, S. M. Brigham, & N. Taber (Eds.), Building on critical traditions: Adult education and learning in Canada<\/i>. Toronto: Thompson Educational Publishing. <\/span><\/span>\n\nBoyd, D. (2010). Privacy and publicity in the context of big data. Paper presented at the 19th International Conference on World Wide Web (WWW2010), 29 April 2010, Raleigh, North Carolina, USA. http:\/\/www.danah.org\/papers\/talks\/2010\/WWW2010.html <\/span><\/span>\n\nBoyd, D., & Crawford, K. (2012). Critical equations for Big Data. Information, Communication & Society, 15<\/i>(5), 662\u2013679. doi:10.1080\/1369118X.2012.678878 <\/span><\/span>\n\nBritish Science Council. (2009). What is science? http:\/\/www.sciencecouncil.org\/definition <\/span><\/span>\n\nCavoukian, A., & Jonas, J. (2012). Privacy by design in the age of big data. https:\/\/privacybydesign.ca\/content\/ uploads\/2012\/06\/pbd-big_data.pdf <\/span><\/span>\n\nCormack, A. (2015). A data protection framework for learning analytics. Community.jisc.ac.uk. http:\/\/bit.ly\/1OdIIKZ <\/span><\/span>\n\nChristopher, J. C., Wendt, D. C., Marecek, J., & Goodman, D. M. (2014) Critical cultural awareness: Contributions to a globalizing psychology. American Psychologist, 69<\/i>, 645\u2013655. http:\/\/dx.doi.org\/10.1037\/a0036851 <\/span><\/span>\n\nDenzin, N., & Lincoln, Y. (Eds.) (2011). The Sage handbook of qualitative research (4th ed.). Thousand Oaks, CA: Sage. <\/span><\/span>\n\nDurand, G., Belacel, N., & Goutte, C. (2015) Evaluation of expert-based Q-matrices predictive quality in matrix factorization models. Proceedings of the 10th European Conference on Technology Enhanced Learning <\/i>(EC-TEL\u201915), 15\u201317 September 2015, Toledo, Spain (pp. 56\u201369). Springer. doi:10.1007%2F978-3-319-24258-3_5<\/span><\/span>\n\nEd Tech Review (2016). Educational Data Mining (EDM). http:\/\/edtechreview.in\/dictionary\/394-what-is-educational-data-mining <\/span><\/span>\n\nEl Emam, K. (1998). The predictive validity criterion for evaluating binary classifiers. Proceedings of the 5th International Software Metrics Symposium <\/i>(Metrics 1998), 20\u201321 November 1998, Bethesda, MD, USA (pp. 235\u2013244). IEEE Computer Society. http:\/\/ieeexplore.ieee.org\/document\/731250\/ <\/span><\/span>\n\nEuropean Data Protection Supervisor. (2015). Leading by example: The EDPS strategy 2015\u20132019. https:\/\/secure.edps.europa.eu\/EDPSWEB\/edps\/site\/mySite\/Strategy2015 <\/span><\/span>\n\nFenwick, T. (2015a). Things that matter in education. In B. Williamson (Ed.), Coding\/learning, software and digital data in education<\/i>. University of Stirling, UK. http:\/\/bit.ly\/1NdHVbw <\/span><\/span>\n\nFenwick, T. (2015b). Professional responsibility in a future of data analytics. In B. Williamson (Ed.) Coding\/ learning, software and digital data in education<\/i>. University of Stirling, UK. http:\/\/bit.ly\/1NdHVbw <\/span><\/span>\n\nFournier, H., & Kop, R. (2015). MOOC learning experience design: Issues and challenges. International Journal on E-Learning, 14<\/i>(3), 289\u2013304. <\/span><\/span>\n\nGergen, J. K., Josselson, R., & Freeman, M. (2015). The promise of qualitative inquiry. American Psychologist, 70<\/i>(1), 1\u20139. <\/span><\/span>\n\nGonzalez-Brenes, J., & Huang, Y. (2015). Your model is predictive \u2014 but is it useful? Theoretical and empirical considerations of a new paradigm for adaptive tutoring evaluation. 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 Education Data Mining <\/i>(EDM2015), 26\u201329 June 2015, Madrid, Spain (pp. 187\u2013194). International Educational Data Mining Society. <\/span><\/span>\n\nGupta, A. (2013). Fraud and misconduct in clinical research: A concern. Perspectives in Clinical Research. 4<\/i>(2), 144\u2013147. doi:10.4103\/2229-3485.111800. <\/span><\/span>\n\nHardt, M. (2014). How big data is unfair: Understanding sources of unfairness in data driven decision making. https:\/\/medium.com\/@mrtz\/how-big-data-is-unfair-9aa544d739de <\/span><\/span>\n\nJones, C., Dirckinck-Holmfeld, L., & Lindstr\u00f6m, B. (2006). A relational, indirect, meso-level approach to CSCL design in the next decade. International Journal of Computer Supported Collaborative Learning, 1<\/i>(1), 35\u201356. <\/span><\/span>\n\nKansanen, P., & Meri, M. (1999). The didactic relation in the teaching\u2013studying\u2013learning process. TNTEE Publications, 2<\/i>, 107\u2013116. http:\/\/www.helsinki.fi\/~pkansane\/Kansanen_Meri.pdf <\/span><\/span>\n\nKitchin, R. (2015). Foreword: Education in code\/space. In B. Williamson (Ed.), Coding\/learning, software and digital data in education<\/i>. University of Stirling, UK. http:\/\/bit.ly\/1NdHVbw <\/span><\/span>\n\nKop, R. (2012). The unexpected connection: Serendipity and human mediation in networked learning. Educational Technology & Society, 15<\/i>(2), 2\u201311. <\/span><\/span>\n\nKop, R., Fournier, H., & Durand, G. (2014). Challenges to research in massive open online courses. Merlot Journal of Online Learning and Teaching, 10<\/i>(1). http:\/\/jolt.merlot.org\/vol10no1\/fournier_0314.pdf <\/span><\/span>\n\nLong, Y., & Aleven, V. (2014). Gamification of joint student\/system control over problem selection in a linear equation tutor. In S. Trausan-Matu, K. E. Boyer, M. Crosby, & K. Panourgia (Eds.), Proceedings of the 12th International Conference on Intelligent Tutoring Systems <\/i>(ITS 2014), 5\u20139 June 2014, Honolulu, HI, USA (pp. 378\u2013387). New York: Springer. doi:10.1007\/978-3-319-07221-0_47 <\/span><\/span>\n\nLong, C., & Siemens, G. (2011, September 12). Penetrating the fog: Analytics in learning and education. EDUCAUSE Review, 46<\/i>(5). http:\/\/er.educause.edu\/articles\/2011\/9\/penetrating-the-fog-analytics-in-learning-and-education <\/span><\/span>\n\nMeyer, J. H. F., & Land, R. (2006). Overcoming barriers to student understanding: Threshold concepts and troublesome knowledge. London: Routledge. <\/span><\/span>\n\nMischel, W. (1968). Personality and assessment<\/i>. Londra: Wiley. <\/span><\/span>\n\nOblinger, D. G. (2012, November 1). Analytics: What we\u2019re hearing. EDUCAUSE Review<\/i>. http:\/\/er.educause.edu\/ articles\/2012\/11\/analytics-what-were-hearingOcumpaugh, J., Baker, R., Gowda, S., Heffernan, N., & Heffernan, C. (2014). Population validity for educational data mining models: A case study in affect detection. British Journal of Educational Technology, 45<\/i>(3), 487\u2013501. <\/span><\/span>\n\nPardo, A., & Siemens, G. (2014). Ethical and privacy principles for learning analytics. British Journal of Educational Technology, 45<\/i>(3), 438\u2013450. <\/span><\/span>\n\nSiemens, G., Dawson, D., & Lynch, L. (2013). Improving the quality and productivity of the higher education sector: Policy and strategy for systems-level deployment of learning analytics. SoLAR. https:\/\/www.itl.usyd.edu.au\/projects\/SoLAR_Report_2014.pdf <\/span><\/span>\n\nStewart, B. (2015). Open to influence: What counts as academic influence in scholarly networked Twitter participation. Learning, Media and Technology, 40<\/i>(3), 287\u2013309. doi:10.1080\/17439884.2015.1015547 <\/span><\/span>\n\nWeller, M. (2011). A pedagogy of abundance. Spanish Journal of Pedagogy, 249<\/i>, 223\u2013236. <\/span><\/span>\n\nWen, M., Yang, D., & Ros\u00e9, C. (2014). 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. 130\u2013137). International Educational Data Mining Society. <\/span><\/span>\n\nXu, D., & Smith Jaggars, S. (2013). Adaptability to online learning: Differences across types of students and academic subject areas. CCRC Working Paper No. 54. Community College Research Center, Teachers College, Columbia University. http:\/\/anitacrawley.net\/Reports\/adaptability-to-online-learning.pdf<\/span><\/a><\/span><\/span><\/span>\n","rendered":"

Rita Kop1<\/sup>, Helene Fournier2<\/sup>, Guillaume Durand2 <\/sup><\/span><\/span><\/p>\n

1<\/sup>E\u011fitim Fak\u00fcltesi, Yorkville \u00dcniversitesi, Birle\u015fik Krall\u0131k<\/span><\/span><\/p>\n

2<\/sup>Bilgi ve \u0130leti\u015fim Teknolojileri, Kanada Ulusal Ara\u015ft\u0131rma Konseyi, Kanada<\/span><\/span><\/p>\n

DOI: 10.18608\/hla17.027<\/span><\/span><\/p>\n

\u00d6Z<\/h2>\n

E\u011fitsel veri madencili\u011fi (EVM) ve \u00f6\u011frenme analiti\u011fi (\u00d6A; Fournier, Kop ve Durand, 2014) hakk\u0131ndaki son yaz\u0131m\u0131zda, nicel ve nitel analiz ara\u00e7lar\u0131n\u0131n kullan\u0131labilirli\u011fi ile ilgili yay\u0131nlar\u0131n hen\u00fcz mevcut olmad\u0131\u011f\u0131 ve \u00f6\u011frenenlere \u00f6z-y\u00f6netimli \u00f6\u011frenme yolculuklar\u0131nda yard\u0131mc\u0131 olabilecek daha fazla ara\u015ft\u0131rman\u0131n yararl\u0131 olaca\u011f\u0131 sonucuna varm\u0131\u015ft\u0131k. Bahsi ge\u00e7en ara\u015ft\u0131rmalardan baz\u0131lar\u0131n\u0131n tekrar edilmesi hayal k\u0131r\u0131kl\u0131\u011f\u0131na yol a\u00e7an sonu\u00e7lar verse de bu yay\u0131nlardan baz\u0131lar\u0131 ger\u00e7ekle\u015ftirilmi\u015ftir. Bu b\u00f6l\u00fcmde, e\u011fitim ve \u00f6\u011frenim ortamlar\u0131n\u0131n sonu\u00e7lar\u0131n\u0131 \u00f6l\u00e7mek ve talep etmek i\u00e7in EVM’nin ve \u00d6A’n\u0131n ge\u00e7erlili\u011fi konusunda ele\u015ftirel bir tutum sergiliyoruz. Ayr\u0131ca, deneysel \u00f6\u011frenme modellerinin yanl\u0131\u015fl\u0131klar\u0131n\u0131 g\u00f6stermek i\u00e7in EVM’nin nas\u0131l kullan\u0131labilece\u011fini de rapor edece\u011fiz. Ara\u015ft\u0131r\u0131lacak di\u011fer boyutlar; \u00f6\u011frenmede insan fakt\u00f6rleri, bu fakt\u00f6rlerin EVM ve \u00d6A ile ili\u015fkileri ve a\u00e7\u0131k \u00f6\u011frenme ortamlar\u0131ndaki ara\u015ft\u0131rmalarda \u201cB\u00fcy\u00fck Veri\u201dyi kullanma eti\u011fidir.<\/span><\/span><\/p>\n

Anahtar Kelimeler<\/span>: E\u011fitsel veri madencili\u011fi (EVM), b\u00fcy\u00fck veri, algoritmalar, rastlant\u0131<\/span><\/span><\/p>\n

Son on y\u0131l i\u00e7inde e\u011fitim ve \u00f6\u011fretim teknolojisi alan\u0131 s\u0131ra d\u0131\u015f\u0131 bir de\u011fi\u015fim ve geli\u015fim g\u00f6stermektedir. E\u011fitim alan\u0131nda yap\u0131lan ge\u00e7mi\u015f ara\u015ft\u0131rmalar; \u00f6\u011frenen, \u00f6\u011freten ve kurs i\u00e7eriklerinden olu\u015fan bir e\u011fitim \u00fc\u00e7geniyle ili\u015fkiliyken (Kansanen ve Meri, 1999; Meyer ve Land, 2006) yeni geli\u015ftirilen teknolojiler \u00f6\u011frenmeyi etkileyen di\u011fer boyutlara vurgu yapmaktad\u0131r. \u00d6rne\u011fin, \u00f6\u011frenme ba\u011flam\u0131 veya \u00f6\u011frenme ortam\u0131 ve kullan\u0131lmakta olan teknolojiler (Bouchard, 2013). Fenwick (2015a), insanlar\u0131n ve kulland\u0131klar\u0131 teknolojilerin birbirinden ba\u011f\u0131ms\u0131z de\u011ferlendirilemeyece\u011fini \u015fu \u015fekilde iddia etmektedir: \u201cMateryal ve sosyal g\u00fc\u00e7ler, e\u011fitim s\u00fcre\u00e7lerinin ve etkinliklerinin ortak olu\u015fumunu nas\u0131l de\u011ferlendirmemiz gerekti\u011fi hususunda \u00f6nemli uygulamalar\u0131 kapsayacak \u015fekilde i\u00e7 i\u00e7edirler\u201d (s. 14). \u0130nsanlar ve teknoloji gibi materyaller aras\u0131nda sadece bir etkile\u015fim de\u011fil, ayn\u0131 zamanda birlikte ya\u015fayan bir ili\u015fki vard\u0131r.<\/span><\/p>\n

Yeni teknolojiler bizi bilgi k\u0131tl\u0131\u011f\u0131 \u00e7a\u011f\u0131ndan, bilgi bollu\u011fu \u00e7a\u011f\u0131na ta\u015f\u0131m\u0131\u015ft\u0131r (Weller, 2011). Art\u0131k sosyal medya, a\u011flar \u00fczerinden k\u00fcresel \u00f6l\u00e7ekte, tu\u011fla duvarlarla s\u0131n\u0131rlanan geleneksel s\u0131n\u0131f d\u0131\u015f\u0131nda ileti\u015fim kurmay\u0131 m\u00fcmk\u00fcn k\u0131l\u0131yor. Bu kadar k\u00fcresel bir \u00f6l\u00e7ekte ileti\u015fim, \u00e7ok uzun zaman \u00f6nce d\u00fc\u015f\u00fcn\u00fclemezdi. Veri ve veri depolama, geli\u015fen teknolojilerin etkisi alt\u0131nda geli\u015fmi\u015ftir. Veri toplamak ve veritaban\u0131na kaydetmek yerine, \u015fimdi algoritmalar ve makine \u00f6\u011frenimi kullan\u0131larak temsil edilip g\u00f6rselle\u015ftirilebilen bulutta depolanan b\u00fcy\u00fck veri ak\u0131\u015flar\u0131yla ilgileniyoruz. Bu veriden \u00f6\u011frenmek i\u00e7in ilgin\u00e7 f\u0131rsatlar sunar, bununla birlikte gizli kavray\u0131\u015flar\u0131 ve ayn\u0131 zamanda \u00f6nemli zorluklar\u0131 ortaya \u00e7\u0131kar\u0131r.<\/span><\/p>\n

E\u011fitim s\u00fcrecindeki payda\u015flar\u0131n -\u00f6\u011frenciler, e\u011fitimciler ve y\u00f6neticiler- t\u00fcm bu bilgi seviyelerini etkili bir \u015fekilde eri\u015febilmelerini, y\u00f6netebilmelerini ve anlamlar\u0131n\u0131 sa\u011flayabilecekleri ile ilgili sorular sorulmu\u015ftur. Bilgisayar bilimciler, tam olarak bunu yapabilen otomatik veri filtreleme ve analizi i\u00e7in f\u0131rsatlar \u00f6nermi\u015ftir: mevcut t\u00fcm verileri elden ge\u00e7irin ve \u00f6\u011frenenlere ba\u011flant\u0131lar sa\u011flay\u0131n ve tercih ettikleri bilgiler, insanlar ve ara\u00e7lar i\u00e7in tavsiyeler ve bunu yaparken \u00f6\u011frenim deneyimini ki\u015fiselle\u015ftirin ve \u00f6\u011frenenlerin \u00f6\u011frenmelerinin y\u00f6netimi ve derinle\u015ftirilmesinde \u00f6\u011frenenlere yard\u0131mc\u0131 olun (Siemens, Dawson ve Lynch, 2013). Ayr\u0131ca, \u00f6\u011frenenlerin etkinliklerinin geride b\u0131rakt\u0131\u011f\u0131 izlerden elde edilen verilere eri\u015filerek b\u00fcy\u00fck kurumsal veri k\u00fcmeleri kullanan ara\u015ft\u0131rma \u00f6rnekleri de ortaya \u00e7\u0131kmaktad\u0131r. (Xu ve Smith Jaggars, 2013).<\/span><\/p>\n

De\u011fi\u015fimlerin tart\u0131\u015f\u0131lmas\u0131nda b\u00fcy\u00fck veriler profesyonel uygulamaya zorlayabilir, \u00f6rne\u011fin Fenwick (2015b), karar verme a\u00e7\u0131s\u0131ndan “bilginin azalt\u0131lmas\u0131n\u0131 vurgulamaktad\u0131r. Veri analiti\u011fi yaz\u0131l\u0131m\u0131 sorunlar\u0131n teknik oldu\u011fu, bilinebilir, \u00f6l\u00e7\u00fclebilir parametrelerden olu\u015ftu\u011funu ve teknik hesaplama yoluyla \u00e7\u00f6z\u00fclebilece\u011fi basit \u00f6nc\u00fclleri ile \u00e7al\u0131\u015f\u0131r. Etik ve de\u011ferlerin karma\u015f\u0131kl\u0131\u011f\u0131, belirsizliklerin ve gerilimlerin, k\u00fclt\u00fcr\u00fcn ve politikalar\u0131n ve hatta verilerin topland\u0131\u011f\u0131 ba\u011flam dikkate al\u0131nmaz \u201d(s. 70). <\/span><\/p>\n

Bu \u00f6nemli bir konudur. Ayr\u0131ca, verileri i\u00e7eren mevcut geli\u015fmeler uyguland\u0131\u011f\u0131nda, g\u00fcnl\u00fck prati\u011fimizi tam olarak anla\u015f\u0131lamayacak \u015fekillerde de\u011fi\u015ftirebilece\u011fini vurgulamaktad\u0131r. \u00d6rne\u011fin, kar\u015f\u0131la\u015ft\u0131rma ve tahminde bir ba\u011f\u0131ml\u0131l\u0131k oldu\u011funda ortaya \u00e7\u0131kan e\u015fitlik sorunlar\u0131na dikkat \u00e7ekmektedir (Fenwick, 2015b). Ayr\u0131ca, onun ara\u015ft\u0131rmas\u0131, potansiyel e\u011fitimcilere ve e\u011fitim ara\u015ft\u0131rmac\u0131lar\u0131na \u00f6\u011fretilen ara\u015ft\u0131rma metodolojilerinin, uygulamalar\u0131n\u0131 geli\u015ftirmek i\u00e7in mevcut b\u00fcy\u00fck veri k\u00fcmeleriyle ba\u015fa \u00e7\u0131kmada tamamen yetersiz oldu\u011fu sonucuna varm\u0131\u015ft\u0131r. Ayr\u0131ca, \u201cprofesyonel eylemlili\u011fin ve hesap verilebilirli\u011fin seviyesini” merak etmektedir. \u00c7ok fazla veri birikimi ve hesaplamas\u0131 otomatikle\u015ftirilmi\u015ftir, bu da algoritmalar\u0131n \u00f6zerkli\u011fi ve k\u00f6t\u00fc \u015feyler oldu\u011funda sorumlulu\u011fu \u00fcstlenmek hakk\u0131nda yeni sorular ortaya \u00e7\u0131karmaktad\u0131r \u201d(s. 71). Bunlar dikkatle d\u00fc\u015f\u00fcn\u00fclmesi gereken ciddi sorulard\u0131r. Bu b\u00f6l\u00fcm, ara\u015ft\u0131rmalarda b\u00fcy\u00fck veri k\u00fcmelerini ve \u00f6\u011frenene destek i\u00e7in kullan\u0131c\u0131 verilerini kullanan e\u011fitsel veri madencili\u011fi ve analiti\u011fi ile ilgili baz\u0131 zorluklar\u0131 ele alacakt\u0131r. Ayn\u0131 zamanda otomasyonun etkisini ve teknolojiyle \u00f6\u011frenmede insan ileti\u015fiminin ve kat\u0131l\u0131m\u0131n\u0131n de\u011fi\u015ftirilmesiyle makinele\u015fmenin olas\u0131 etkilerini de ara\u015ft\u0131racakt\u0131r.<\/span><\/p>\n

\u00d6\u011eRENME Y\u00d6NET\u0130M\u0130NDE E\u011e\u0130TSEL VER\u0130 MADENC\u0130L\u0130\u011e\u0130 FIRSATLARI<\/h2>\n

G\u00fcvenilirlik ve Ge\u00e7erlilik<\/h3>\n

E\u011fitsel veri madencili\u011fi (EVM), e\u011fitim ortamlar\u0131ndan gelen benzersiz veri t\u00fcrlerini ara\u015ft\u0131rmak i\u00e7in y\u00f6ntemler geli\u015ftirmekle ve \u00f6\u011frencilerin ve \u00f6\u011frendikleri ortamlar\u0131 daha iyi anlamak i\u00e7in bu y\u00f6ntemleri kullanmakla ilgilenen bir disiplindir (Ed Tech Review, 2016). E\u011fitsel veri madencili\u011fi, ad\u0131ndan da daha geni\u015f bir kapsam\u0131 ifade etmekte olup bilgi getirimi ve \u00f6\u011frenme mekanizmalar\u0131n\u0131n daha iyi anla\u015f\u0131lmas\u0131 i\u00e7in e\u011fitim verilerinin ara\u015ft\u0131r\u0131lmas\u0131ndan da \u00f6te bir anlam\u0131 ifade etmektedir. Bu nedenle, EVM ayn\u0131 zamanda ge\u00e7erlilik, tekrar \u00fcretilebilirlik ve genelle\u015ftirilebilirlik ile ilgili bilimsel kayg\u0131larla y\u00f6netilen istatistiksel yakla\u015f\u0131mlar\u0131 kullanarak makine \u00f6\u011frenmesini ve istatistiksel yakla\u015f\u0131mlar\u0131 kullanarak \u00f6\u011frenen davran\u0131\u015flar\u0131n\u0131 tahmin etmek i\u00e7in y\u00f6ntemler ve modeller geli\u015ftirmeyi ama\u00e7lamaktad\u0131r.<\/span><\/p>\n

\u00d6\u011frenme analiti\u011fi (\u00d6A), EVM alan\u0131yla yak\u0131ndan ilgilidir ve \u00f6\u011frenmeyi ve ger\u00e7ekle\u015fti\u011fi ortamlar\u0131 anlamak ve optimize etmek amac\u0131yla \u00f6\u011frenenler ve ba\u011flamlar\u0131 hakk\u0131ndaki verilerin \u00f6l\u00e7\u00fcm\u00fc, toplanmas\u0131, analizi ve raporlanmas\u0131 ile ilgilidir (Long ve Siemens, 2011). \u00d6\u011frenme s\u00fcrecini artt\u0131rmak i\u00e7in EVM teknikleri ve \u00d6A kullan\u0131l\u0131r. Bunlar etkili \u00f6\u011frenci deste\u011fi sa\u011flanmas\u0131na yard\u0131mc\u0131 olma konusunda \u00fcmit verici g\u00f6r\u00fcnmektedir ve bu yeni geli\u015fmelerin e\u011fitimi ve \u00f6\u011frenmeyi art\u0131raca\u011f\u0131 vaadine ra\u011fmen, \u00f6nemli zorluklar da tespit edilmi\u015ftir.<\/span><\/p>\n

Bir \u00f6l\u00e7\u00fcde, EVM sadece d\u00fcnyan\u0131n d\u00f6rt bir yan\u0131ndan ara\u015ft\u0131rmac\u0131lar\u0131n \u00fcretken katk\u0131lar\u0131ndan dolay\u0131 geli\u015fen bir ara\u015ft\u0131rma alan\u0131 de\u011fil ayn\u0131 zamanda bir bilimdir. Son zamanlarda, \u0130ngiltere Bilim Konseyi bilimi \u201ckan\u0131tlara dayal\u0131 sistematik bir metodoloji izleyerek do\u011fal ve sosyal d\u00fcnyan\u0131n bilgi ve anlay\u0131\u015f\u0131 pe\u015finde olmak\u201d olarak tan\u0131mlam\u0131\u015ft\u0131r (British Science Council, 2009). Kan\u0131t, di\u011fer herhangi bir bilimsel alanda oldu\u011fu gibi alanda yap\u0131lan herhangi bir iddia i\u00e7in bir gerekliliktir; e\u011fitsel veri madencili\u011fi ve analiti\u011fi ara\u015ft\u0131rmac\u0131lar\u0131, e\u011fitim verilerinden al\u0131nan veya do\u011frulanan iddialar\u0131 ve ke\u015fifleri desteklemek veya reddetmek i\u00e7in kan\u0131ta ihtiya\u00e7 duyar.<\/span><\/p>\n

Bununla birlikte, sa\u011flam ya da zay\u0131f kan\u0131t\u0131n ne oldu\u011funa dair ortak bir tan\u0131mlama, \u201cpozitif\u201d (Bilgisayar Bilimi) ve \u201csosyal\u201d bilimlerden (E\u011fitim) bir\u00e7ok bilim insan\u0131n\u0131 bir araya getiren EVM ve \u00d6A ara\u015ft\u0131rma toplulu\u011fu i\u00e7in a\u00e7\u0131k de\u011fildir. Burada, kendi ara\u015ft\u0131rmam\u0131z s\u0131ras\u0131nda rastlad\u0131\u011f\u0131m\u0131z baz\u0131 tutars\u0131zl\u0131klar ve y\u00f6ntemsel kusurlardan \u00f6rnekler sunaca\u011f\u0131z. Veri payla\u015f\u0131m\u0131 sayesinde, Long ve Aleven (2014) bir ak\u0131ll\u0131 \u00f6\u011fretici sisteminde oyunla\u015ft\u0131r\u0131lm\u0131\u015f bir yakla\u015f\u0131m\u0131n \u00f6\u011frenme iddialar\u0131yla ters d\u00fc\u015febildiler. Ancak bazen veri k\u00fcmelerini payla\u015fmak yeterli de\u011fildir; baz\u0131 ara\u015ft\u0131rma \u00e7al\u0131\u015fmalar\u0131, bir ara\u015ft\u0131rma belgesinde a\u00e7\u0131k\u00e7a tan\u0131mlanmas\u0131 zor olabilecek ad\u0131mlar esnas\u0131nda birden \u00e7ok se\u00e7im (yanl\u0131l\u0131k) yap\u0131ld\u0131\u011f\u0131 i\u00e7in kapsaml\u0131 bir \u00f6n i\u015flemeyi gerektirir. Bu y\u00fczden ayn\u0131 kurallar\u0131 izleyerek veri k\u00fcmesini haz\u0131rlamaya \u00e7al\u0131\u015fan ba\u015fka bir ekip bunu yapmay\u0131 ba\u015faramayabilir. \u00d6n i\u015fleme s\u0131ras\u0131nda kullan\u0131lan yaz\u0131l\u0131m\u0131n niteli\u011fi de etkili olabilir. \u00d6nemli y\u00f6ntemlerin uygulanmas\u0131, R, SPSS, Matlab ve di\u011fer ara\u00e7lar\u0131 kullan\u0131rken de\u011fi\u015fiklik g\u00f6stererek potansiyel olarak farkl\u0131 \u00e7\u0131kar\u0131mlara yol a\u00e7abilir.<\/span><\/p>\n

Bir ba\u015fka ihtilafl\u0131 y\u00f6n, istatistiksel bir modelin \u00fczerine kurulaca\u011f\u0131 deney \u00f6ncesi varsay\u0131mlar<\/i><\/span> veya \u201ctemel referans de\u011fer\u201d dir. \u00d6rne\u011fin, insan uzmanlar\u0131n maddeler ve beceri e\u015flemelerini tan\u0131mlad\u0131\u011f\u0131 yeterlik \u00e7er\u00e7eveleri sorgulanabilirdir. (Durand, Belacel ve Goutte, 2015). Di\u011fer bir\u00e7ok gizli \u00f6zellik gibi becerileri nitelendirmek bazen zordur. Bu ama\u00e7la, PSLC Veri Ma\u011fazas\u0131 \u00f6\u011frenme uzmanlar\u0131na, \u00f6\u011frencilerden edinilen g\u00f6zlem sonu\u00e7lar\u0131n\u0131n yapt\u0131klar\u0131 e\u015flemeleri geli\u015ftirmeye ve payla\u015fmaya yard\u0131mc\u0131 oldu\u011fundan yeterlilik \u00e7er\u00e7evelerini test etmeleri i\u00e7in inan\u0131lmaz bir ortam sunar. Ayr\u0131ca, veri k\u00fcmelerini payla\u015fmak sorunlar\u0131 tan\u0131mlama konusunda de\u011ferli oldu\u011fundan, veri k\u00fcmelerini payla\u015fmak EVM ve \u00d6A pratisyenleri aras\u0131nda harika bir ara\u00e7t\u0131r. Payla\u015f\u0131m ola\u011fan olmal\u0131 ve yay\u0131nlanan hi\u00e7bir sonu\u00e7 di\u011fer ekiplerin bu iddialar\u0131 do\u011frulama imk\u00e2n\u0131 olmadan ciddi \u015fekilde d\u00fc\u015f\u00fcn\u00fclmemelidir.<\/span><\/p>\n

Di\u011fer baz\u0131 konular, istatistiksel \u00e7al\u0131\u015fmalar\u0131 ve \u00f6zellikle do\u011frusal korelasyon \u00f6l\u00e7\u00fclerini g\u00f6z \u00f6n\u00fcnde bulundurmak gibi sorular do\u011furabilir. EVM, neyin kan\u0131t olarak kullan\u0131labilece\u011fi veya kullan\u0131lamayaca\u011f\u0131 konusunda farkl\u0131 uygulamalar\u0131 olan ve farkl\u0131 bak\u0131\u015f a\u00e7\u0131lar\u0131na sahip ara\u015ft\u0131rmac\u0131lar\u0131 bir araya getirir. \u201cPozitif bilimlerde\u201d, r =.5’in alt\u0131ndaki \u00f6nemli bir Pearson korelasyonunun sistematik olarak zay\u0131f oldu\u011fu d\u00fc\u015f\u00fcn\u00fcl\u00fcrken, \u201csosyal bilimlerde\u201d, r =.3 de\u011ferlerinin g\u00fc\u00e7l\u00fc oldu\u011funu d\u00fc\u015f\u00fcnmek normaldir. Psikologlar bile buna.3 e\u015fi\u011fi, \u201cki\u015filik katsay\u0131s\u0131\u201d diyorlar \u00e7\u00fcnk\u00fc ki\u015filik \u00f6zellikleri ile davran\u0131\u015flar aras\u0131ndaki ili\u015fkilerin \u00e7o\u011fu, yetkinlik ve performans aras\u0131ndaki ili\u015fki de d\u00e2hil olmak \u00fczere, bu de\u011fer etraf\u0131nda olma e\u011filimindedir (Mischel, 1968, s. 78). EVM’de duygu analizi ile ilgili yap\u0131lan \u00e7al\u0131\u015fmalar (Wen, Yang ve Rose, 2014), KA\u00c7D’lerde okulu b\u0131rakma oranlar\u0131 konusunda anlaml\u0131 olan \u201csosyal\u201d bilim ara\u015ft\u0131rma sonu\u00e7lar\u0131ndan bilgi i\u015flemsel sonu\u00e7lar elde etmenin zor oldu\u011fu bir \u00f6rnek sunmaktad\u0131r. Ayr\u0131ca, incelenmekte olan konunun nitel tekniklerle daha iyi ara\u015ft\u0131r\u0131lmas\u0131 \u00f6nerilebilir. Bununla birlikte, ama\u00e7 \u00e7\u0131kar\u0131mlarda bulunmak oldu\u011funda nicel EVM formundaki ili\u015fkiler zay\u0131f kalmaktad\u0131r. \u00d6zellikle, tan\u0131m gere\u011fi kriterdeki varyans\u0131n %9’unu a\u00e7\u0131klayan bir a., 3. korelasyonu, duygu analizi alan\u0131ndaki tahminlerde s\u0131n\u0131rl\u0131 bir de\u011fere sahip olabilir.<\/span><\/p>\n

Birka\u00e7 istatistik testi de anlaml\u0131 olabilir ancak sonu\u00e7lar\u0131n kesinli\u011fi konusunda tam olarak ger\u00e7e\u011fe uygun olmayabilir. El Emam (1998), bir Ki-Kare testinin bir s\u0131n\u0131fland\u0131r\u0131c\u0131n\u0131n \u00f6ng\u00f6r\u00fcc\u00fc ge\u00e7erlili\u011fini de\u011ferlendirmede nas\u0131l yan\u0131lt\u0131c\u0131 olabilece\u011fini de\u011ferlendirmi\u015ftir. Daha yak\u0131n zamanlarda, Gonzalez-Brenes ve Huang (2015), Leopard \u00f6l\u00e7\u00fcm\u00fcn\u00fc uyarlanabilir \u00f6zel ders sistemlerini de\u011ferlendirmenin ve onlar\u0131n faydal\u0131l\u0131klar\u0131n\u0131 de\u011ferlendirerek sistemin \u00f6ng\u00f6r\u00fcc\u00fc do\u011frulu\u011funu de\u011ferlendirme sonu\u00e7lar\u0131n\u0131 art\u0131rman\u0131n standart bir yolu olarak \u00f6nermi\u015ftir. Onlar bu sistemlerdeki \u00f6\u011frenme \u00e7\u0131kt\u0131lar\u0131na eri\u015fmek i\u00e7in \u00f6\u011frenenlerden istenecek \u00e7aba miktar\u0131n\u0131 de\u011ferlendirmeyi teklif ettiler. Sonu\u00e7ta, fayda \u00f6l\u00e7\u00fcleri, sistemleri kullanan ki\u015filerin en \u00e7ok ilgilendikleri \u015fey olabilir.<\/span><\/p>\n

Bu ama\u00e7la, EVM toplulu\u011funun \u00f6nde gelen ara\u015ft\u0131rmac\u0131lar\u0131ndan Ryan Baker, KA\u00c7D\u2019de \u201cE\u011fitimde B\u00fcy\u00fck Veri\u201d ba\u015fl\u0131kl\u0131 ara\u015ft\u0131rmas\u0131nda, ara\u015ft\u0131rmac\u0131lar\u0131n modellerinin ge\u00e7erlili\u011fini anlamalar\u0131na ve daha ak\u0131lc\u0131 bir \u015fekilde kontrol etmelerine yard\u0131mc\u0131 olacak \u00f6rnekler ve iyi uygulama \u00f6nerileri sunmaktad\u0131r (genelle\u015ftirilebilirlik, ekolojik, yap\u0131 ve tahmine dayal\u0131, esas ve i\u00e7erik ge\u00e7erlili\u011fi). Baker bu dersinde, kesinlik, ROC (e\u011frilik yar\u0131\u00e7ap\u0131), keskinlik ve hassasiyet veya Ki-Kare gibi s\u0131n\u0131fland\u0131r\u0131c\u0131lara ili\u015fkin di\u011fer \u00f6l\u00e7\u00fclerin kusurlar\u0131n\u0131 bertaraf etmede nas\u0131l \u201cduyargan\u0131n \u015fanstan daha iyi oldu\u011funu\u201d ve belli bir \u00f6zelli\u011fi \u201cduyargan\u0131n do\u011fru bir \u015fekilde tespit etme olas\u0131l\u0131\u011f\u0131n\u0131\u201d \u00f6l\u00e7mek i\u00e7in s\u0131ras\u0131yla Kappa’y\u0131 ve hatta daha iyi olan A’y\u0131 kulland\u0131\u011f\u0131n\u0131 vurgulad\u0131. (Ocumpaugh, Baker, Gowda, Heffernan ve Heffernan, 2014, s. 492). Ancak EVM yay\u0131nlar\u0131nda A\u2019 ve Kappa kullan\u0131m\u0131 \u015fu ana kadar s\u0131n\u0131rl\u0131 g\u00f6r\u00fcnmektedir.<\/span><\/p>\n

Ara\u015ft\u0131rma b\u00fct\u00fcnl\u00fc\u011f\u00fcn\u00fcn son derece \u00f6nemli oldu\u011funu vurgulamak istiyoruz. EVM ve \u00d6A’da \u201cuydurulmu\u015f\u201d sonu\u00e7lar vermek \u00e7ekici olabilir. Kendi \u00e7al\u0131\u015fmam\u0131z, somut sonu\u00e7lar\u0131n elde edilmesinin genellikle bir\u00e7ok \u00e7aba gerektirdi\u011fini, bir\u00e7ok \u00e7al\u0131\u015fman\u0131n ba\u015far\u0131 garantisi olmadan yap\u0131ld\u0131\u011f\u0131n\u0131 ve do\u011frulama s\u00fcrecinin sorunlu g\u00f6r\u00fcnd\u00fc\u011f\u00fcn\u00fc g\u00f6stermi\u015ftir. Bug\u00fcne kadar ger\u00e7e\u011fi \u00e7arp\u0131tacak \u00f6nemli durumlar ortaya \u00e7\u0131kmam\u0131\u015ft\u0131r ancak di\u011fer bilimsel alanlarda g\u00f6zlemlendi\u011fi gibi gelecekteki olas\u0131 sahtek\u00e2rl\u0131k ve suistimal davalar\u0131ndan ka\u00e7\u0131nmak i\u00e7in \u015feffafl\u0131\u011fa ili\u015fkin y\u00f6nergeler sunma konusu daha fazla dikkate al\u0131nmal\u0131d\u0131r (Gupta, 2013).<\/span><\/p>\n

\u00d6A ve EVM’nin geli\u015fmekte olan alanlar\u0131nda, bilimsel hedefin iddial\u0131 oldu\u011funu ve ara\u015ft\u0131rmac\u0131lar\u0131n iddialar\u0131n\u0131n bilimsel sa\u011flaml\u0131\u011f\u0131n\u0131 dikkatlice kontrol etmelerini gerektirdi\u011fini savunuyoruz. Ara\u015ft\u0131rmalar yak\u0131n gelecekte insan \u00f6\u011frenmesini etkileyecekleri s\u00f6z\u00fcne dair verilen ba\u011f\u0131\u015flar ile desteklense de \u00d6A ve EVM alanlar\u0131n\u0131n bilimsel d\u00fcr\u00fcstl\u00fc\u011f\u00fc korumak i\u00e7in zaman ay\u0131rmalar\u0131 \u00f6nemlidir. Bu kullan\u0131lan y\u00f6ntemlerin ve elde edilen sonu\u00e7lar\u0131n taraf\u0131m\u0131zca dikkatle de\u011ferlendirilmesini gerektirir.<\/span><\/p>\n

Nitel Veri Analizinin Zorluklar\u0131<\/h3>\n

E\u011fitim ara\u015ft\u0131rmalar\u0131n\u0131n son on y\u0131ldaki geli\u015fimine bakacak olursak, nicelden nitel ara\u015ft\u0131rmalara do\u011fru belirgin bir kay\u0131\u015f vard\u0131r (Gergen, Josselson ve Freeman, 2015). Psikologlar, \u00f6\u011frenme ve bilmenin karma\u015f\u0131kl\u0131\u011f\u0131n, bireylerin davran\u0131\u015flar\u0131n\u0131 tek ba\u015f\u0131na test ederek belirleyemedikleri fikrini giderek daha fazla desteklemektedir. \u00d6\u011frenenlerin i\u00e7inde ya\u015fad\u0131klar\u0131 topluma dair eylemlerinin ve d\u00fc\u015f\u00fcncelerinin zenginli\u011fi ve bilgi a\u011flar\u0131nda yer alan ki\u015filerle ileti\u015fim kurma \u00e7al\u0131\u015fmalar\u0131, insanlar\u0131n bilgi geli\u015ftirme ve \u00f6\u011frenmeleri hakk\u0131nda daha derin, daha kapsay\u0131c\u0131 ve son derece k\u00fclt\u00fcrel bir anlay\u0131\u015f sa\u011flar (Christopher, Wendt, Marecek ve Goodman, 2014; Denzin ve Lincoln, 2011; Gergen vd., 2015).<\/span><\/p>\n

Mevcut teknoloji y\u00f6n\u00fcnden zengin \u00f6\u011frenme ortam\u0131, sadece duvarl\u0131 bir s\u0131n\u0131f de\u011fildir, ayn\u0131 zamanda k\u00fcresel a\u011f ileti\u015fimini de i\u00e7erir; bunlar, ara\u015ft\u0131rmac\u0131lara ara\u015ft\u0131rma metodolojilerini yeniden icat etmeleri konusunda zorlayan, d\u00f6n\u00fc\u015fl\u00fc hik\u00e2yeleme ve zengin imgeleri kapsar. T\u00fcm bunlar\u0131n \u00f6tesine ge\u00e7mek -\u00f6\u011frencilerin memnuniyetini ortaya \u00e7\u0131karmak i\u00e7in yap\u0131lan ders sonu anketler, \u00f6\u011frenenler taraf\u0131ndan \u00fcretilen verileri ara\u015ft\u0131r\u0131lmas\u0131, \u00e7evrimi\u00e7i \u00f6\u011frenme deneyimi s\u0131ras\u0131nda \u00fcretilen anlat\u0131lar, g\u00f6r\u00fcnt\u00fcler ve g\u00f6rselle\u015ftirmeleri analiz etmek- \u00f6\u011frenme etkile\u015fimlerinin zengin dokusunu anlamak i\u00e7in se\u00e7enekler sunar. Art\u0131k \u00f6\u011frenme ortam\u0131n\u0131n bir par\u00e7as\u0131n\u0131 olu\u015fturan sosyal medya da de\u011fi\u015fen kelime ve resim gruplar\u0131ndaki temel boyutlar\u0131 analiz etmek, \u00f6\u011frenme s\u00fcrecinin kalbine resm\u00ee ders de\u011ferlendirmelerinden \u00e7ok daha fazla girebilir.<\/span><\/p>\n

\u0130ki kitlesel a\u00e7\u0131k \u00e7evrimi\u00e7i dersi (KA\u00c7D), PLENK2010 ve CLOM REL 2014 ara\u015ft\u0131rmas\u0131, bu ara\u015ft\u0131rman\u0131n i\u00e7erdi\u011fi zorluklar\u0131n alt\u0131n\u0131 \u00e7izmektedir (Fournier ve Kop, 2015; Kop, Fournier ve Durand, 2014). \u00d6nceki KA\u00c7D ara\u015ft\u0131rmas\u0131, verilerdeki kal\u0131plar\u0131 g\u00f6rselle\u015ftirmek i\u00e7in g\u00fc\u00e7l\u00fc ara\u00e7lar\u0131, \u00f6zellikle de dijital sosyal a\u011flarda her zamankinden daha b\u00fcy\u00fck ve daha zengin veri k\u00fcmeleri sa\u011flad\u0131. Bununla birlikte, bu t\u00fcr kal\u0131plar\u0131 a\u00e7\u0131\u011fa \u00e7\u0131karma \u00e7al\u0131\u015fmalar\u0131, verilerin \u00fcretildi\u011fi pedagojik ve teknik ba\u011flamdaki cevaplardan daha fazla soru sa\u011flam\u0131\u015ft\u0131r. KA\u00c7D kat\u0131l\u0131mc\u0131lar\u0131n\u0131n neden yapt\u0131klar\u0131 verileri \u00fcrettiklerini anlamaya \u00e7al\u0131\u015f\u0131rken nitel bir yakla\u015f\u0131ma do\u011fru ilerlemek, b\u00fcy\u00fck veri, EVM ve \u00d6A’n\u0131n bize karma\u015f\u0131k \u00f6\u011frenme s\u00fcre\u00e7leri ve deneyimleri hakk\u0131nda neler s\u00f6yleyebilece\u011fi ve s\u00f6yleyemedi\u011fi<\/i><\/span> \u00fczerine ele\u015ftirel bir derin d\u00fc\u015f\u00fcnmeye neden oldu.<\/span><\/p>\n

Boyd (2010) bunu \u015f\u00f6yle ifade etmi\u015ftir:<\/span><\/p>\n

B\u00fcy\u00fck Veri’yi \u00e7evreleyen co\u015fkunun \u00e7o\u011fu, bir parma\u011f\u0131n\u0131z\u0131 t\u0131klatarak b\u00fcy\u00fck miktarda veriye kolayca eri\u015febilmekten kaynaklan\u0131yor. Ya da Vint Cerf in ifadesiyle, \u201c\u0130nsanl\u0131k tarihinde hi\u00e7bir zaman, bu kadar \u00e7abuk ve bu kadar kolay bilgiye ula\u015famad\u0131k\u201d Ne yaz\u0131k ki, bu heyecanda kaybolan \u015fey, bu verilerin ne oldu\u011funun ve ne anlama geldi\u011finin ele\u015ftirel bir analizidir. (s. 2).<\/div>\n

Bu kadar \u00e7ok veri ve bilgi ile \u00e7ok h\u0131zl\u0131 bir \u015fekilde u\u011fra\u015f\u0131rken, ara\u015ft\u0131rmac\u0131lar\u0131n verileri, hedef kitle taraf\u0131ndan faydal\u0131 ve eri\u015filebilir bir bi\u00e7imde sindirebilecekleri anla\u015f\u0131labilir, t\u00fcketilebilir veya i\u015flem yap\u0131labilir sunum bi\u00e7imlerine d\u00f6n\u00fc\u015ft\u00fcrmek i\u00e7in en uygun s\u00fcre\u00e7leri ve teknikleri tasarlamalar\u0131 gerekir. Karma\u015f\u0131k fikirleri etkili bir \u015fekilde ula\u015ft\u0131rma yetene\u011fi, ara\u015ft\u0131rma bulgular\u0131n\u0131 prati\u011fe d\u00f6n\u00fc\u015ft\u00fcren de\u011ferli bir \u015fey \u00fcretmede kritik \u00f6neme sahiptir. E\u011fitim s\u00fcrecindeki payda\u015flar\u0131n (\u00f6r. \u00d6\u011frenciler, e\u011fitimciler ve y\u00f6neticiler) t\u00fcm bu bilgi seviyelerine nas\u0131l etkili bir \u015fekilde eri\u015febilecekleri, y\u00f6netebilecekleri ve anlam \u00e7\u0131karacaklar\u0131 ile ilgili sorular sorulmu\u015ftur; EVM ve \u00d6A y\u00f6ntemleri, tam olarak otomatik veri filtreleme ve analizin bunu nas\u0131l yapabilece\u011fine i\u015faret etmektedir. Bu \u00f6\u011frenme ve \u00f6\u011frenenler hakk\u0131nda potansiyel olarak zengin \u00e7\u0131kar\u0131mlara yol a\u00e7abilir ancak ayn\u0131 zamanda s\u00fcre\u00e7te bir\u00e7ok yeni ilgin\u00e7 ara\u015ft\u0131rma sorusu ve zorlu\u011fu da beraberinde getirebilir. Bunu yaparken, ara\u015ft\u0131rmac\u0131lar verinin ne kadar anlaml\u0131 oldu\u011funu g\u00f6stermenin yan\u0131 s\u0131ra, duyarl\u0131 ara\u015ft\u0131rma tasar\u0131mlar\u0131 ve uygulamalar\u0131 ile sorumlu inovasyon ile me\u015fgul olarak e\u011fitim s\u00fcrecinde \u00e7e\u015fitli payda\u015flara hitap etmek i\u00e7in \u00e7aba g\u00f6stermelidirler (Berland, Baker ve Blikstein, 2014).<\/span><\/p>\n

Algoritmalar, Mutlu Tesad\u00fcf ve \u00d6\u011frenmede \u201c\u0130nsan\u201d: \u00d6\u011frenme Analiti\u011fine Ele\u015ftirel Bir Bak\u0131\u015f<\/span><\/span><\/h4>\n

EVM ve \u00d6A’da yava\u015f yava\u015f bir alan yaz\u0131n geli\u015fmektedir. Temel olarak, \u00f6\u011frenmeyi analiz etmek i\u00e7in teknolojiyi kullanmak ya da \u00f6\u011frenmeyi ilerletmek i\u00e7in yorday\u0131c\u0131 analiti\u011fi kullanmak kolay de\u011fildir. E\u011fitimde algoritmalar\u0131n ve di\u011fer veri g\u00fcd\u00fcml\u00fc sistemlerin geli\u015ftirilmesine ili\u015fkin konular, bu sistemlerin ger\u00e7ekte neleri de\u011fi\u015ftirdi\u011fi ve bu de\u011fi\u015fimin olumlu mu olumsuz mu oldu\u011fu ile ilgili sorulara yol a\u00e7ar. \u0130kincisi, veri g\u00fcd\u00fcml\u00fc sistemlerin i\u00e7eri\u011fini kim etkiliyor ve e\u011fitim s\u00fcrecine ne gibi katk\u0131lar sa\u011flayabilir?<\/span><\/p>\n

\u00c7evrimi\u00e7i e\u011fitimde ancak ayn\u0131 zamanda ba\u011fda\u015ft\u0131r\u0131c\u0131 bir a\u011f ortam\u0131nda (Jones, Dirckinck \u2013 Holmfeld ve Lindstrom, 2006), \u00f6\u011frenme u\u011fra\u015f\u0131ndaki kat\u0131l\u0131mc\u0131lar aras\u0131ndaki ileti\u015fim ve diyalog kaliteli bir \u00f6\u011frenme deneyiminin merkezinde olmu\u015ftur. Bu insan dokunu\u015fu \u00f6\u011frenme sistemleri ve ortamlar\u0131n\u0131 geli\u015ftirmede gerekli bir bile\u015fendir (Bates, 2014). Bilgili di\u011fer ki\u015filerin mevcudiyeti ve kat\u0131l\u0131m\u0131, kat\u0131l\u0131mc\u0131lar\u0131n resm\u00ee \u00f6\u011frenme ortamlar\u0131nda ancak ayn\u0131 zamanda \u00e7evrimi\u00e7i ilgili a\u011flarda da fikirlerini, yarat\u0131c\u0131l\u0131klar\u0131n\u0131 ve d\u00fc\u015f\u00fcncelerini geni\u015fletmek i\u00e7in her zaman hayati olarak g\u00f6r\u00fclm\u00fc\u015ft\u00fcr (Jones vd., 2006).<\/span><\/p>\n

\u00d6\u011frenme i\u00e7in veri g\u00fcd\u00fcml\u00fc teknolojiler geli\u015ftirirken, bu insan unsurunu bir \u015fekilde \u00f6\u011frenme s\u00fcrecinin yarar\u0131 i\u00e7in kullanmak \u00f6nemlidir. Bu bilginin filtrelenmesinde veya Sokratik sorular\u0131n sorulmas\u0131nda, bilginin toplanmas\u0131na insan arac\u0131l\u0131\u011f\u0131yla arac\u0131l\u0131k edilmesi gerekti\u011fi anlam\u0131na gelir (Kop, 2012). Kaynaklar hakk\u0131nda bilgi ve ba\u011flant\u0131lar sa\u011flayan \u201ctakip\u00e7iler\u201din kullan\u0131c\u0131 taraf\u0131ndan se\u00e7ildi\u011fi, de\u011ferli ve g\u00fcvenilir olarak g\u00f6r\u00fcld\u00fc\u011f\u00fc Twitter gibi sosyal mikroblog sitelerinin bunu ba\u015far\u0131l\u0131 bir \u015fekilde yapt\u0131klar\u0131 g\u00f6sterilmi\u015ftir (Bista, 2014; Kop, 2012; Stewart, 2015). Algoritmalarda, bu kararlar\u0131n elde edilmesi zordur ancak belki de verilere dayanan \u00f6neri sistemleri ile, ileti\u015fimi temel alan bili\u015fsel destek ve \u00f6\u011frenme deste\u011fi uygulamalar\u0131n\u0131n bir kombinasyonu bunu kolayla\u015ft\u0131rabilir.<\/span><\/p>\n

Veri odakl\u0131 sistemlerin i\u00e7eri\u011fini kimin etkiledi\u011fini ve e\u011fitim s\u00fcrecine ne gibi bir de\u011fer katabileceklerini d\u00fc\u015f\u00fcnmek \u00f6nemlidir. Ayr\u0131ca, sadece yeni teknolojilerin sundu\u011fu f\u0131rsatlar ve verimlili\u011fi dikkate al\u0131nmakla kalmay\u0131p, ayn\u0131 zamanda, insan ileti\u015fimi ile betimlenen bir \u00f6\u011frenme ortam\u0131ndan, \u00fczerinde \u00f6\u011frenenin kontrol\u00fcn\u00fcn az oldu\u011fu veya hi\u00e7 kontrol\u00fcn\u00fcn olmad\u0131\u011f\u0131 teknik unsurlar\u0131 i\u00e7eren bir ortama ge\u00e7meye ili\u015fkin etik unsurlar da d\u00fc\u015f\u00fcn\u00fclmelidir.<\/span><\/p>\n

Algoritmalar\u0131n geli\u015ftirilmesinde \u00f6ne s\u00fcr\u00fclen sorunlardan biri, \u00f6neri veya arama sonucunun niteli\u011fini etkileyebilecek olan ara\u015ft\u0131rmac\u0131n\u0131n araca dair yanl\u0131l\u0131\u011f\u0131n\u0131n tan\u0131t\u0131lmas\u0131d\u0131r. (Hardt, 2014). Verilerle \u00e7al\u0131\u015fan ki\u015filerin do\u011fru \u015fekilde e\u011fitilmesi, her \u015feyi de\u011fi\u015ftirebilir (Fenwick, 2015b; Boyd ve Crawford, 2012). Halen uygulamalar\u0131 sosyal bilimlerde bir ge\u00e7mi\u015fe sahip olmayan bilgisayar bilimcileri ve matematik\u00e7iler \u00fcretmektedir. Boyd ve Crawford’un (2012) ikna edici bir \u015fekilde ileri s\u00fcrd\u00fc\u011f\u00fc gibi:<\/span><\/p>\n

Bilgi i\u015flemsel beceriler en de\u011ferli olarak konumland\u0131r\u0131ld\u0131\u011f\u0131nda, kimlerin avantajl\u0131 oldu\u011fu ve b\u00f6yle bir ba\u011flamda kimin dezavantajl\u0131 oldu\u011fu konusunda sorular ortaya \u00e7\u0131kar. Bu, bilgisayar bilimcilerinin ve sosyal bilimcilerin ikisinin de \u00f6nerecekleri de\u011ferli bak\u0131\u015f a\u00e7\u0131lar\u0131na sahip olduklar\u0131n\u0131 kabul etmek yerine \u201csay\u0131lar\u0131 okuyabilen\u201dler etraf\u0131nda kendince yeni hiyerar\u015filer olu\u015fturur. Bu \u00f6nemli \u00f6l\u00e7\u00fcde, ayn\u0131 zamanda cinsiyetlendirilmi\u015f bir b\u00f6l\u00fcnmedir. \u015eu anda bilgi i\u015flemsel beceriye sahip ara\u015ft\u0131rmac\u0131lar\u0131n \u00e7o\u011fu erkektir ve feminist tarih\u00e7iler ve bilim felsefecilerinin g\u00f6sterdi\u011fi gibi, sorular\u0131 kimlerin sordu\u011fu hangi sorular\u0131n sorulaca\u011f\u0131n\u0131 belirler (s.674).<\/div>\n

Boyd ve Crawford (2012) bilgisayar bilimcileri ve sosyal bilimcilerin \u00f6nyarg\u0131s\u0131z, y\u00fcksek nitelikli analiz ara\u00e7lar\u0131 geli\u015ftirmek i\u00e7in birlikte \u00e7al\u0131\u015fmas\u0131 gerekti\u011fini ve farkl\u0131 alanlardaki insanlarla ekip \u00e7al\u0131\u015fmas\u0131n\u0131n b\u00fcy\u00fck veri madencili\u011fi ve analizinde de verimli olabilece\u011fini \u00f6ne s\u00fcr\u00fcyor. Tabii ki, verilerin b\u00fcy\u00fcmesi ve kullan\u0131labilirli\u011fi de bunlardan faydalanmay\u0131 \u00e7ekici k\u0131lm\u0131\u015ft\u0131r ancak yine de baz\u0131 zorluklar vard\u0131r. \u00c7o\u011fu zaman insanlar bilgileri g\u00fcvendikleri kaynaklardan al\u0131rlar ancak Fenwick’in (2015b) \u00f6nerdi\u011fi gibi, yeni usullerin kullan\u0131m\u0131 \u201cg\u00fcnl\u00fck uygulama ve sorumluluklar\u0131 tam olarak tan\u0131namayacak \u015fekillerde\u201d de\u011fi\u015ftirebilir (s. 71). \u00d6rne\u011fin o, kar\u015f\u0131la\u015ft\u0131rma ve \u00f6ng\u00f6rmeye g\u00fcvenmenin b\u00fcy\u00fck veriler dikkatle kullan\u0131lmad\u0131\u011f\u0131nda \u00f6zellikle algoritmalar\u0131 \u00fcreten insanlar kal\u0131pla\u015fm\u0131\u015f kan\u0131lar\u0131n peki\u015ftirildi\u011finin fark\u0131nda de\u011fillerse, \u201ckendi kendini peki\u015ftirebilece\u011fi ve \u00fcretebilece\u011fi, s\u00fcre\u00e7 ba\u011f\u0131ml\u0131l\u0131\u011f\u0131n\u0131 artt\u0131rabilece\u011fi ve mevcut e\u015fitsizlikleri hapsedebilece\u011fini\u201d vurguluyor.<\/span><\/p>\n

Ayr\u0131ca, h\u00e2lihaz\u0131rda kullan\u0131lmakta olan algoritmalar\u0131n \u00e7o\u011funun daha derin \u00f6\u011frenme seviyelerini artt\u0131rmak veya topluma de\u011fer katmak i\u00e7in de\u011fil ekonomik kazan\u0131m i\u00e7in \u00fcretildi\u011fini hafife almamal\u0131y\u0131z. Kitchin (2015) taraf\u0131ndan iddia edildi\u011fi gibi, \u201cYaz\u0131l\u0131m yaln\u0131zca bir dizi talimat\u0131 yerine getiren bir kod sat\u0131r\u0131 de\u011fildir, bir\u00e7ok ak\u0131l\u0131n farkl\u0131 sosyal, politik ve ekonomik ili\u015fkiler i\u00e7erisine konumlanm\u0131\u015f sonucu, ko\u015fullu, ili\u015fkisel ve ba\u011flamsal olarak ortaya \u00e7\u0131kan bir sosyal \u00fcr\u00fcn\u00fc olarak anla\u015f\u0131lmal\u0131d\u0131r. \u201d(s. 5). A\u00e7\u0131k\u00e7as\u0131 otomatikle\u015ftirilmi\u015f algoritma sistemlerinin geli\u015ftirilmesinin, bir \u015feyler ters gitti\u011finde kimin sorumlu oldu\u011funu i\u015faret etmenin zor olabilece\u011fi ba\u015fka bir do\u011fal problemi bulunmaktad\u0131r.<\/span><\/p>\n

Baz\u0131 Etik Hususlar<\/h3>\n

A\u00e7\u0131k \u00f6\u011frenme ortamlar\u0131, g\u00fc\u00e7l\u00fc veri analizi ara\u00e7lar\u0131 ve y\u00f6ntemleri ile bir araya getirildi\u011finde, yeni ili\u015fkiler ve \u00f6\u011frenmeye destek sa\u011flaman\u0131n yan\u0131 s\u0131ra, \u00f6\u011frenenleri insan ileti\u015fimi ile nitelendirilen bir ortamdan, \u00f6\u011freneni \u00fczerinde \u00e7ok az ya da hi\u00e7 kontrol\u00fc olmad\u0131\u011f\u0131 teknik unsurlar i\u00e7eren bir ortama ta\u015f\u0131yan \u00f6nemli etik sorunlar\u0131 ve zorluklar\u0131 da vurgulamaktad\u0131r. Genel a\u011f geli\u015ftirme konusundaki ticari \u00e7aban\u0131n \u00e7o\u011fu b\u00fcy\u00fck veriden beslenmektedir ve herhangi bir yenilik\u00e7i e\u011fitim anlay\u0131\u015f\u0131ndan yoksundur (Atkinson, 2015). Biz \u201cTeknoloji \u00e7\u00f6z\u00fcmlerinin, e\u011fitime kayda de\u011fer ve s\u00fcrd\u00fcr\u00fclebilir faydalar getirmesini sa\u011flayacak anlaml\u0131 pedagojik ve yeti\u015fkin \u00f6\u011frenme teorilerine dayanan, uzmanl\u0131k ve ara\u015ft\u0131rmalar taraf\u0131ndan bilgilendirilmi\u015f \u00f6\u011frenme tasar\u0131m\u0131n\u0131n kendisi oldu\u011funu” kabul ediyoruz (Atkinson, 2015, s. 7).<\/span><\/p>\n

EVM ve \u00d6A da d\u00e2hil olmak \u00fczere teknolojik yenili\u011fin dinamik h\u0131z\u0131 ayn\u0131 zamanda gizlili\u011fin de\u011fi\u015fime sebep olabilecek bir eyleme ge\u00e7me bi\u00e7imiyle korunmas\u0131n\u0131 gerektirir. Bu amaca ula\u015fmak i\u00e7in, EVM ve ileri analitik alan\u0131ndaki ara\u015ft\u0131rmac\u0131lar ve sistem tasar\u0131mc\u0131lar\u0131, mahremiyet g\u00fc\u00e7lendirici teknolojileri do\u011frudan \u00fcr\u00fcn ve s\u00fcre\u00e7lerine entegre eden sorumlu yenilikleri uygulamak zorundad\u0131r (Cavoukian ve Jonas, 2012). Oblinger (2012)’e g\u00f6re, \u201cAnalitik k\u00fclt\u00fcr meselesidir, bir sorgulama k\u00fclt\u00fcr\u00fcd\u00fcr: soru sormak, veri aramak, verilerin a\u00e7\u0131\u011fa vurdu\u011fu g\u00fc\u00e7l\u00fc ve zay\u0131f y\u00f6nlere kar\u015f\u0131 d\u00fcr\u00fcst olmak, bu \u00e7abalar\u0131n sonu\u00e7lar\u0131 meyvesini verirken uyum sa\u011flamakt\u0131r.\u201d(s. 98).<\/span><\/p>\n

Bunu ak\u0131lda tutarak, yeni nesil analitikleri tasarlayan ve geli\u015ftirenlerin Tasar\u0131m yoluyla Mahremiyet konusunda bilgilendirilmelerini sa\u011flaman\u0131z\u0131 \u015fiddetle tavsiye ederiz. Bu hesap verebilirlik, ara\u015ft\u0131rma do\u011frulu\u011fu, veri koruma, mahremiyet ve r\u0131zay\u0131 i\u00e7eren fark\u0131ndal\u0131k ve sorumlu uygulamay\u0131 gerektirir (Cavoukian ve Jonas, 2012; Cormack, 2015). \u00d6zel ve halka a\u00e7\u0131k veriler aras\u0131ndaki \u00e7izgi, daha fazla a\u00e7\u0131k \u00f6\u011frenme ortamlar\u0131na kat\u0131l\u0131m f\u0131rsat\u0131 yarat\u0131ld\u0131\u011f\u0131ndan ve kat\u0131l\u0131mc\u0131lar, onlar\u0131n etkinlikleri ve onlar\u0131n davran\u0131\u015flar\u0131 Facebook, Twitter, Google gibi sosyal medya ve di\u011fer \u00e7evrim i\u00e7i kullan\u0131labilir potansiyel sosyal medya ara\u00e7lar\u0131 yoluyla eri\u015filebilir oldu\u011fundan giderek bulan\u0131kla\u015fmaktad\u0131r. B\u00fcy\u00fck veri ba\u011flam\u0131nda biz, \u201c\u0130nsanlar algoritmalar\u0131n kendileri ile ilgili nas\u0131l ili\u015fkiler ve varsay\u0131mlar yaratabileceklerini ve birle\u015ftirilmi\u015f ki\u015fisel bilginin onlar\u0131n davran\u0131\u015flar\u0131 ile ilgili izinsiz ve m\u00fcdahaleci h\u00fck\u00fcmlere nas\u0131l d\u00f6nebilece\u011fini anlamak istiyor” (s.10) \u015feklinde g\u00f6r\u00fc\u015f belirten Avrupa Veri Koruma Denet\u00e7isine (2015) kat\u0131l\u0131yoruz.<\/span><\/p>\n

SONU\u00c7<\/h2>\n

B\u00fcy\u00fck veri \u00e7al\u0131\u015fmalar\u0131nda do\u011fruluk, kontrol, \u015feffafl\u0131k ve g\u00fc\u00e7 ile ilgili \u00f6nemli sorular\u0131n da ele al\u0131nmas\u0131 gerekir. Pardo ve Siemens (2014), \u00e7ok uzun s\u00fcre \u00e7ok fazla veri saklaman\u0131n (\u00f6\u011frenci dijital verileri, mahremiyete duyarl\u0131 veriler d\u00e2hil) zarar verebilece\u011fini ve ki\u015fisel verileri korumak i\u00e7in emanet edilen sisteme veya kuruma g\u00fcvensizli\u011fine yol a\u00e7abilece\u011fini belirtmektedir. B\u00fcy\u00fck veri eti\u011fi konusundaki tart\u0131\u015fmalar, veri temizli\u011fi, veri se\u00e7imi ve yorumlanmas\u0131 (Boyd ve Crawford, 2012), veri analiti\u011finin istilac\u0131l\u0131k olas\u0131l\u0131\u011f\u0131 ve insan ileti\u015fiminin ve otomatik makine \u00f6\u011frenme algoritmalar\u0131 ve geri bildirimleri le ili\u015fkisinin olas\u0131 makinele\u015ftirme etkilerine ili\u015fkin metodolojik kayg\u0131lara vurgu yapm\u0131\u015ft\u0131r. Ara\u015ft\u0131rmac\u0131lar ve geli\u015ftiricilerin (Fenwick, 2015b) yararl\u0131 gelecek ad\u0131mlar\u0131 in\u015fa etmek i\u00e7in (veri madencili\u011fi ve ak\u0131ll\u0131 \u00f6\u011frenme analitikleri d\u00e2hil olmak \u00fczere) b\u00fcy\u00fck verinin sundu\u011fu imk\u00e2n ve s\u0131n\u0131rl\u0131l\u0131klar konusunda dikkatli olmalar\u0131 gerekir. Ara\u015ft\u0131rmac\u0131lar, \u00e7al\u0131\u015fmalar\u0131nda do\u011fabilecek yanl\u0131\u015f ve yanl\u0131l\u0131klar\u0131n bir k\u0131sm\u0131ndan ka\u00e7\u0131nmak ve e\u011fitim s\u00fcrecine de\u011fer katmak i\u00e7in b\u00fcy\u00fck veri ve veri g\u00fcd\u00fcml\u00fc sistemlerdeki \u00f6nemli sorunlar\u0131 ve zorluklar\u0131 ele almak i\u00e7in ekipler halinde birlikte \u00e7al\u0131\u015fmal\u0131d\u0131r.<\/span><\/p>\n

KAYNAK\u00c7A<\/h2>\n

Atkinson, S. P. (2015). Adaptive learning and learning analytics: A new learning design paradigm. BPP Working paper. https:\/\/spatkinson.files.wordpress.com\/2015\/05\/atkinson-adaptive-learning-and-learning-analytics.pdf <\/span><\/span><\/p>\n

Bates, T. (2014). Two design models for online collaborative learning: Same or different? http:\/\/www.tonybates.ca\/2014\/11\/28\/two-design-models-for-online-collaborative-learning-same-or-different\/ <\/span><\/span><\/p>\n

Berland, M., Baker, R. S., & Blikstein, P. (2014). Educational data mining and learning analytics: Applications to constructionist research. Technology, Knowledge and Learning, 19<\/i>, 206\u2013220. <\/span><\/span><\/p>\n

Bista, K. (2014). Is Twitter an effective pedagogical tool in higher education? Perspectives of education graduate students. Journal of the Scholarship of Teaching and Learning, 15<\/i>(2), 83\u2013102. http:\/\/files.eric.ed.gov\/ fulltext\/EJ1059422.pdf <\/span><\/span><\/p>\n

Bouchard, P. (2013). Education without a distance: Networked learning. In T. Nesbit, S. M. Brigham, & N. Taber (Eds.), Building on critical traditions: Adult education and learning in Canada<\/i>. Toronto: Thompson Educational Publishing. <\/span><\/span><\/p>\n

Boyd, D. (2010). Privacy and publicity in the context of big data. Paper presented at the 19th International Conference on World Wide Web (WWW2010), 29 April 2010, Raleigh, North Carolina, USA. http:\/\/www.danah.org\/papers\/talks\/2010\/WWW2010.html <\/span><\/span><\/p>\n

Boyd, D., & Crawford, K. (2012). Critical equations for Big Data. Information, Communication & Society, 15<\/i>(5), 662\u2013679. doi:10.1080\/1369118X.2012.678878 <\/span><\/span><\/p>\n

British Science Council. (2009). What is science? http:\/\/www.sciencecouncil.org\/definition <\/span><\/span><\/p>\n

Cavoukian, A., & Jonas, J. (2012). Privacy by design in the age of big data. https:\/\/privacybydesign.ca\/content\/ uploads\/2012\/06\/pbd-big_data.pdf <\/span><\/span><\/p>\n

Cormack, A. (2015). A data protection framework for learning analytics. Community.jisc.ac.uk. http:\/\/bit.ly\/1OdIIKZ <\/span><\/span><\/p>\n

Christopher, J. C., Wendt, D. C., Marecek, J., & Goodman, D. M. (2014) Critical cultural awareness: Contributions to a globalizing psychology. American Psychologist, 69<\/i>, 645\u2013655. http:\/\/dx.doi.org\/10.1037\/a0036851 <\/span><\/span><\/p>\n

Denzin, N., & Lincoln, Y. (Eds.) (2011). The Sage handbook of qualitative research (4th ed.). Thousand Oaks, CA: Sage. <\/span><\/span><\/p>\n

Durand, G., Belacel, N., & Goutte, C. (2015) Evaluation of expert-based Q-matrices predictive quality in matrix factorization models. Proceedings of the 10th European Conference on Technology Enhanced Learning <\/i>(EC-TEL\u201915), 15\u201317 September 2015, Toledo, Spain (pp. 56\u201369). Springer. doi:10.1007%2F978-3-319-24258-3_5<\/span><\/span><\/p>\n

Ed Tech Review (2016). Educational Data Mining (EDM). http:\/\/edtechreview.in\/dictionary\/394-what-is-educational-data-mining <\/span><\/span><\/p>\n

El Emam, K. (1998). The predictive validity criterion for evaluating binary classifiers. Proceedings of the 5th International Software Metrics Symposium <\/i>(Metrics 1998), 20\u201321 November 1998, Bethesda, MD, USA (pp. 235\u2013244). IEEE Computer Society. http:\/\/ieeexplore.ieee.org\/document\/731250\/ <\/span><\/span><\/p>\n

European Data Protection Supervisor. (2015). Leading by example: The EDPS strategy 2015\u20132019. https:\/\/secure.edps.europa.eu\/EDPSWEB\/edps\/site\/mySite\/Strategy2015 <\/span><\/span><\/p>\n

Fenwick, T. (2015a). Things that matter in education. In B. Williamson (Ed.), Coding\/learning, software and digital data in education<\/i>. University of Stirling, UK. http:\/\/bit.ly\/1NdHVbw <\/span><\/span><\/p>\n

Fenwick, T. (2015b). Professional responsibility in a future of data analytics. In B. Williamson (Ed.) Coding\/ learning, software and digital data in education<\/i>. University of Stirling, UK. http:\/\/bit.ly\/1NdHVbw <\/span><\/span><\/p>\n

Fournier, H., & Kop, R. (2015). MOOC learning experience design: Issues and challenges. International Journal on E-Learning, 14<\/i>(3), 289\u2013304. <\/span><\/span><\/p>\n

Gergen, J. K., Josselson, R., & Freeman, M. (2015). The promise of qualitative inquiry. American Psychologist, 70<\/i>(1), 1\u20139. <\/span><\/span><\/p>\n

Gonzalez-Brenes, J., & Huang, Y. (2015). Your model is predictive \u2014 but is it useful? Theoretical and empirical considerations of a new paradigm for adaptive tutoring evaluation. 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 Education Data Mining <\/i>(EDM2015), 26\u201329 June 2015, Madrid, Spain (pp. 187\u2013194). International Educational Data Mining Society. <\/span><\/span><\/p>\n

Gupta, A. (2013). Fraud and misconduct in clinical research: A concern. Perspectives in Clinical Research. 4<\/i>(2), 144\u2013147. doi:10.4103\/2229-3485.111800. <\/span><\/span><\/p>\n

Hardt, M. (2014). How big data is unfair: Understanding sources of unfairness in data driven decision making. https:\/\/medium.com\/@mrtz\/how-big-data-is-unfair-9aa544d739de <\/span><\/span><\/p>\n

Jones, C., Dirckinck-Holmfeld, L., & Lindstr\u00f6m, B. (2006). A relational, indirect, meso-level approach to CSCL design in the next decade. International Journal of Computer Supported Collaborative Learning, 1<\/i>(1), 35\u201356. <\/span><\/span><\/p>\n

Kansanen, P., & Meri, M. (1999). The didactic relation in the teaching\u2013studying\u2013learning process. TNTEE Publications, 2<\/i>, 107\u2013116. http:\/\/www.helsinki.fi\/~pkansane\/Kansanen_Meri.pdf <\/span><\/span><\/p>\n

Kitchin, R. (2015). Foreword: Education in code\/space. In B. Williamson (Ed.), Coding\/learning, software and digital data in education<\/i>. University of Stirling, UK. http:\/\/bit.ly\/1NdHVbw <\/span><\/span><\/p>\n

Kop, R. (2012). The unexpected connection: Serendipity and human mediation in networked learning. Educational Technology & Society, 15<\/i>(2), 2\u201311. <\/span><\/span><\/p>\n

Kop, R., Fournier, H., & Durand, G. (2014). Challenges to research in massive open online courses. Merlot Journal of Online Learning and Teaching, 10<\/i>(1). http:\/\/jolt.merlot.org\/vol10no1\/fournier_0314.pdf <\/span><\/span><\/p>\n

Long, Y., & Aleven, V. (2014). Gamification of joint student\/system control over problem selection in a linear equation tutor. In S. Trausan-Matu, K. E. Boyer, M. Crosby, & K. Panourgia (Eds.), Proceedings of the 12th International Conference on Intelligent Tutoring Systems <\/i>(ITS 2014), 5\u20139 June 2014, Honolulu, HI, USA (pp. 378\u2013387). New York: Springer. doi:10.1007\/978-3-319-07221-0_47 <\/span><\/span><\/p>\n

Long, C., & Siemens, G. (2011, September 12). Penetrating the fog: Analytics in learning and education. EDUCAUSE Review, 46<\/i>(5). http:\/\/er.educause.edu\/articles\/2011\/9\/penetrating-the-fog-analytics-in-learning-and-education <\/span><\/span><\/p>\n

Meyer, J. H. F., & Land, R. (2006). Overcoming barriers to student understanding: Threshold concepts and troublesome knowledge. London: Routledge. <\/span><\/span><\/p>\n

Mischel, W. (1968). Personality and assessment<\/i>. Londra: Wiley. <\/span><\/span><\/p>\n

Oblinger, D. G. (2012, November 1). Analytics: What we\u2019re hearing. EDUCAUSE Review<\/i>. http:\/\/er.educause.edu\/ articles\/2012\/11\/analytics-what-were-hearingOcumpaugh, J., Baker, R., Gowda, S., Heffernan, N., & Heffernan, C. (2014). Population validity for educational data mining models: A case study in affect detection. British Journal of Educational Technology, 45<\/i>(3), 487\u2013501. <\/span><\/span><\/p>\n

Pardo, A., & Siemens, G. (2014). Ethical and privacy principles for learning analytics. British Journal of Educational Technology, 45<\/i>(3), 438\u2013450. <\/span><\/span><\/p>\n

Siemens, G., Dawson, D., & Lynch, L. (2013). Improving the quality and productivity of the higher education sector: Policy and strategy for systems-level deployment of learning analytics. SoLAR. https:\/\/www.itl.usyd.edu.au\/projects\/SoLAR_Report_2014.pdf <\/span><\/span><\/p>\n

Stewart, B. (2015). Open to influence: What counts as academic influence in scholarly networked Twitter participation. Learning, Media and Technology, 40<\/i>(3), 287\u2013309. doi:10.1080\/17439884.2015.1015547 <\/span><\/span><\/p>\n

Weller, M. (2011). A pedagogy of abundance. Spanish Journal of Pedagogy, 249<\/i>, 223\u2013236. <\/span><\/span><\/p>\n

Wen, M., Yang, D., & Ros\u00e9, C. (2014). 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. 130\u2013137). International Educational Data Mining Society. <\/span><\/span><\/p>\n

Xu, D., & Smith Jaggars, S. (2013). Adaptability to online learning: Differences across types of students and academic subject areas. CCRC Working Paper No. 54. Community College Research Center, Teachers College, Columbia University. http:\/\/anitacrawley.net\/Reports\/adaptability-to-online-learning.pdf<\/span><\/a><\/span><\/span><\/span><\/p>\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-126","chapter","type-chapter","status-publish","hentry","chapter-type-numberless"],"part":109,"_links":{"self":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters\/126","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\/126\/revisions"}],"part":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/parts\/109"}],"metadata":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapters\/126\/metadata\/"}],"wp:attachment":[{"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/media?parent=126"}],"wp:term":[{"taxonomy":"chapter-type","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/pressbooks\/v2\/chapter-type?post=126"},{"taxonomy":"contributor","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/contributor?post=126"},{"taxonomy":"license","embeddable":true,"href":"https:\/\/acikkitap.com.tr\/oaek\/wp-json\/wp\/v2\/license?post=126"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}