應用大數據分析開發適性化教材推薦系統;Applying Big Data Analytics to Develop an Adaptive Course Material Recommendation System

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/72068

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/72068

Title:	應用大數據分析開發適性化教材推薦系統;Applying Big Data Analytics to Develop an Adaptive Course Material Recommendation System
Authors:	王傑生;WANG,CHIEH-SHENG
Contributors:	資訊工程學系
Keywords:	開放教育資源;推薦系統;協同過濾;Clickstream
Date:	2016-07-22
Issue Date:	2016-10-13 14:24:27 (UTC+8)
Publisher:	國立中央大學
Abstract:	近年來開放教育資源(Open Educational Resources)的興起，允許使用者開放修改及分享OER 教材並提供教師免費或是低成本的教學資源以降低教師提供課程的門檻，提供更多讓學習者受教的機會。然而開放式教育資源發展至今，教材數目越來越龐大，導致使用者在為數眾多的教材中尋找所需的教材時需要花費大量的時間及人力。本研究所應用的開放教育資源平台同樣也遭遇相同的問題，透過將推薦系統導入該平台雖然可以有效的減輕教材數目所帶來的影響，然而現有的推薦系統使用內容導向式演算法產生與教材內容相關的教材推薦雖然可以推薦相關教材，但是卻缺少使用者回饋，修正教材推薦的結果使推薦結果能更加貼近使用者。因此，本研究透過蒐集使用者在開放教育資源的瀏覽歷程(Click-stream)作為推測使用者喜好的參考，為了處理大量的使用者瀏覽歷程資料，本研究使用Spark 資料分析框架作為開發工具，將使用者依照喜好相似程度透過分群演算法建立使用者族群模型，最後結合瀏覽歷程及使用者族群模型推測使用者的喜好並給予適性化的教材推薦，使教材推薦結果更符合使用者需求且幫助使用者縮短搜尋教材的時間。;Open Educational Resources has become popular for serval years. It allows user to share and revise the open educational resources. By providing no or low cost educational resources to institutions and educators brings more opportunity to the students to access to education. However, with developed for many years the amount of open educational resources has become more and more huge, it costs a lot of time for users to find the right educational resource and it becomes difficult for users to find the educational resources that they really want. This paper find that an open educational resources platform has the same problem. Although it already has a content-based recommendation system, the lack of user feedbacks leads to it cannot make the recommendation result fit to the user requirement. In this paper, we collect the click-stream from user as the preference of user. In order to deal with vast amount of data, we use Spark to build user cluster model by user’s preference. Finally, we combined current user’s preference with user cluster model and recommended the adaptive educational resource to current user.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	248	View/Open

社群 sharing

Loading...