Book Recommender System for Wikipedia Article Readers in a University Library

摘要

Wikipedia has emerged as an important source of information for university students. It has been reported that university students tend to read and do research with Wikipedia articles more than they do with books, even when within a library. To encourage students to read library books as a more reliable source of information, a library system was developed for recommending library books to Wikipedia readers within a particular university library. The proposed system assigns a Nippon Decimal Classification (NDC) category to each Wikipedia article and recommends library books in the same NDC category to readers of the article. In the test implementation of the system, the precision of assigning NDC categories to Wikipedia articles using a convolutional neural network was as high as 87.4%, while the precision of selecting books for recommendation using a support vector machine reached 99.8%.