基於探勘和群眾分包法之課程與網路資源推薦系統

Recommending Web Resources to Course Units Based on Mining and Crowdsourcing Approaches

郭宏哲(Hung-Che Kuo)

指導教授 ： 林宣華

國立暨南國際大學/科技學院/資訊工程學系/碩士(2013年)

摘要

隨著資訊科技的進步，網際網路的普及與人們使用網路的頻率越來越高，漸漸的網路上各種專業領域知識的網站越來越多，網頁的資訊量也以爆炸性的成長，現在人們往往遇到一些需要解決問題的時候，皆會上網找尋適當的答案，網路已然成為快速且及時獲得知識的平台。藉由搜尋引擎搜尋得到回傳的內容，充斥著各式各樣類型的資訊，造成了多數使用者無法快速的找到自己需要的答案，即使透過搜尋引擎找尋資訊，其龐大的資訊卻讓人不知所措，還得自己整理所需的資訊，過程實在消耗過多的人力與時間。若我們可以將同性質之學習使用者做分類，將可準確且快速的推薦使用者學習上所需資源內容。
在本論文中以「九年一貫之國中小教育」領域做研究，由於其主題及課程架構太過於複雜，使用者對於不同的課程與單元很難有效率的找到適合用於教學的網路資源。因此，本論文提出以Meta Search及群眾外包 (Crowdsourcing) 技術，提供個人化教學資源推薦服務，讓使用者能夠更快速的取得個人化用於教學的教學資源。首先，為了實作Crowdsourcing的機制，於Google Chrome瀏覽器擴充功能中建置了Edu 2.0 Category App，讓領域專家在瀏覽器看到適合資源時，可以如同網路書籤服務，立刻將該支援推薦到合適的領域、年級、出版社教科書版本，更進一步將資源推薦到相關的課程單元中。這樣的服務方便各領域專家及對於教育有熱忱的使用者，即時推薦資源。系統也開發透明化與彈性化的記錄機制，記錄使用者的瀏覽操作過程，並分析使用者個人化的需求，達到教學資源自動化的推薦。最後，系統分析Crowdsourcing資料，以改進Edu 2.0系統，提升推薦網路教學資源的品質，並以個人化服務改善平台使用者的服務品質。

關鍵字

Meta Search ； 資料探勘 ； 群眾外包

並列摘要

With the development of the Internet information technology, people reply on the Web to learn and share knowledge. The various domain knowledge is available from the Web through facilities supported by search engines, social sites and professional sites. However, the explosive growth of information makes people hard to obtain desired information or knowledge so that the Web is far from the e-learning environment, especially in the educational domain.
In this thesis, based on the portal directory constructed from “the General Guidelines of Grades 1-9 Curriculum for Elementary and Junior High School Education”, we propose the crowdsourcing system that provides Chrome Extension App for users. While users (teachers, volunteers, parent, students, or guest) encounter a great website or page through Chrome browser, they can add the useful learning material into the G9 Curriculum. By browsing the G9 directories, they can recommend the material to adequate grades, domains, or units. Based on the crowdsourcing mechanism provided by the thesis, the system transparently logs the data through the Chrome App. Then, we propose the method to analyze the quality and contribution of each users so that the system can categorize them into user or domain expert. Experiments show that the proposed method is able to log and analyze crowd data and recognize users who are domain experts.

並列關鍵字

Meta Search ； Data Mining ； Crowdsourcing

參考文獻

[1]Boris Chidlovskii, Natalie S. Glance and M. Antonietta Grasso, “Collaborative Re-Ranking of Search Results,” In The National Conference on Artificial Intelligence 2000 Workshop on AI for Web Search, pages 18-23, 2000.

[2]Crowdsourcing, “Crowdsourcing - Wikipedia,” http://en.wikipedia.org/wiki/Crowdsourcing.

[3]Danial E. Dreilinger, “Description and Evaluation of a Meta-Search Agent,” The degree of Master of Science Colorado State University, pages 4-12, 1996.

[4]Edu 2.0 Category課程單元類目共享服務, https://chrome.google.com/webstore/detail/edu-20-category/mkjbngkcodknlmfacodkkbaklfndajkj.

[5]Enhanced Entity–Relationship Model, “EER - Wikipedia,” http://en.wikipedia.org/wiki/Enhanced_entity%E2%80%93relationship_model.