運用課程大綱資訊以建置圖書館外文圖書推薦服務

The Development of A Library Book Recommendation Service based on Course Syllabi

莊翊鈞(Yi-Chun Chuang)

指導教授 ： 林福仁

國立清華大學/科技管理學院/服務科學研究所/碩士(2019年)

摘要

圖書館服務的價值主要在於獲取的內容的流通，學生會透過搜尋欄位輸入關鍵字來搜尋書籍，以獲得與搜尋的關鍵字相關的書籍列表，如果學生沒有一定程度的領域知識，便可能無法輸入精確的關鍵字來獲取相關書籍。對於圖書館來說，如何促進學生充分利用可用的資源是圖書館最優先的服務。因此，這樣對圖書館及學生是互惠互利的，雙方都可以共同合作來達成學習的目標。
除了促進學生獲取相關的內容之外，大學圖書館旨在擴大學生的知識範圍，以實現大學的教育目標，培養跨領域人才。從大學圖書館取得館藏書籍資料，以及從學校教務處取得課程資訊來為學生發展圖書推薦服務，讓學生可以借閱與修習課程有相關的書籍以拓展知識，這樣會是個有用的試驗。
因此在本研究中，我們從國立清華大學圖書館取得書籍資料，以及從教務處所提供的課程總表搜集課程大綱資訊，來建立圖書推薦服務。我們提出的方法包含兩個部份，第一部份是分別擷取課程大綱以及書籍資料的關鍵字，接著運用TF-IDF演算法來建立關鍵字矩陣，最後透過計算餘弦相似性（Cosine Similarity）來匹配關鍵字以推薦出書籍；第二部份是推薦服務的運作，系統會根據學生所選擇的課程來推薦書籍列表。最後，透過描述性統計以及計算推薦的精確度，來評測我們所發展的系統。
藉由成功地執行與測試我們所發展的圖書館書籍推薦服務，我們預計學生將可以取得與他們所修習的課程的相關書籍，進而增加圖書館的書籍流通量。

關鍵字

文件探勘 ； 推薦系統 ； 圖書館 ； 基於內容的推薦

並列摘要

The value of library service mainly lies on the circulation of contents acquired. Students are used to search books by issuing keywords via search bar to obtain a list of books relevant to the queried keywords. Students without certain levels of domain knowledge may not be able to issue precise keywords to obtain relevant books. For libraries, how to facilitate students to fully utilize the resources available is the highest priority of library service. Thus, it is a mutually beneficial for a library and students that both can collaborate to achieve the goal.
Besides facilitating students to obtain relevant contents, a university library is aimed to expand students’ knowledge spectrum to achieve the educational objectives of the university to cultivate interdisciplinary talents. Taking the data of book collection in a university library and courses information from the Office of Academic Affairs, it is a potentially useful trial to develop a recommendation system for students to borrow books relevant to their courses by which to expand their knowledge spectrum.
In this research, we took the book collection data from the library of National Tsing Hua University and course syllabi from the Office of Academic Affairs to build a library book recommendation system. The proposed method consists of two parts. Part 1 is to extract keywords of books and syllabi, respectively, and create document-term matrices by using term frequency–inverse document frequency (tf–idf), and then use cosine similarity to match the keywords to recommend the books. Part 2 is the operation of recommendation. The system will recommend a list of books according to a student’s courses they choose. Finally, we evaluate the system by using descriptive statistics and calculating the precision.
By successfully implementing and testing the proposed book recommendation system for library service, we anticipate students obtain the book relative to their courses, which in turn increases the circulation of library books.

並列關鍵字

Text mining ； Recommendation system ； library ； content-based recommendation

參考文獻

Balabanović, M., & Shoham, Y. (1997). Fab: content-based, collaborative recommendation. Communications of the ACM, 40(3), 66-72.

Basu, C., Hirsh, H., & Cohen, W. (1998). Recommendation as classification: Using social and content-based information in recommendation. Paper presented at the Aaai/iaai.

Chen, H.-L. (2007). PORE : A Personal Ontology Recommender System for NCHU Library.

Davis, J., & Goadrich, M. (2006). The relationship between Precision-Recall and ROC curves. Paper presented at the Proceedings of the 23rd international conference on Machine learning.

Huang, A. (2008). Similarity measures for text document clustering. Paper presented at the Proceedings of the sixth new zealand computer science research student conference (NZCSRSC2008), Christchurch, New Zealand.