CAJ | 학술논문

个性化图书推荐已成为图书馆领域关注的热点问题，但面临着读者兴趣、图书内容难以获取以及“冷启动”等一系列挑战。文中基于图书借阅行为建立“读者—图书”二部图模型，并基于此提出个性化图书推荐方法。该方法首先根据书名计算图书之间相似度；其次，基于读者兴趣相似度对读者进行聚类，并生成每个读者的获选图书集合；最后计算每个读者与候选图书集合中每本图书的匹配度，并排序后输出推荐图书列表。实验结果表明，该方法能在未知读者兴趣、图书内容的情况下，有效地实现个性化图书推荐，并缓解了“冷启动”问题。
개성화도서추천이성위도서관영역관주적열점문제，단면림착독자흥취、도서내용난이획취이급“랭계동”등일계렬도전。문중기우도서차열행위건립“독자—도서”이부도모형，병기우차제출개성화도서추천방법。해방법수선근거서명계산도서지간상사도；기차，기우독자흥취상사도대독자진행취류，병생성매개독자적획선도서집합；최후계산매개독자여후선도서집합중매본도서적필배도，병배서후수출추천도서렬표。실험결과표명，해방법능재미지독자흥취、도서내용적정황하，유효지실현개성화도서추천，병완해료“랭계동”문제。
Personalized book recommendations have become a hot area in library science. Current recommending methods,however,are facing the difficulty to automatically acquire reader interests and book topics,and the “cold start” problem. A novel personalized book recommending method based on “Reader-Book” bipartite graph derived from the book lending behavior is proposed. First,the semantic similarities among books are calculated utilizing the book titles. Second,readers are divided into different groups with the use of clustering analysis based on the similarity of reader interests. Every group is assigned a selected book set. Finally,each reader is recommended a preferable book list based on the matching degree between reader and book. Experimental results show that this method can recommend personalized books to a reader without knowing reader interests and book topics,and alleviate the “cold start” problem.