利用關聯規則發掘讀者適性化之書籍推薦

Abstract

本論文以讀者之借閱資料為探勘的資料來源，每一筆借閱資料包含讀者曾經借閱的書籍，並以某一讀者為探勘的目標，利用關聯規則(Association rules)分別從以下兩方面發掘此讀者適性化的書籍推薦：一是探勘關聯規則，其前置項目組必須包含於此讀者的借閱資料中，根據關聯規則所顯示出的傾向特徵，做為發掘此讀者適性化之書籍推薦的依據；二是在考量讀者借閱興趣的加權情況下，探勘興趣加權關聯規則(Association rules with weighted interest-items)，其前置項目組必須包含於此讀者的借閱資料中，根據興趣加權關聯規則所顯示出的傾向特徵，做為發掘此讀者適性化之書籍推薦的依據。根據文中所提出的探勘方法，我們設計與建置一個發掘讀者個人適性化的書籍推薦系統。此探勘結果，對圖書館在擬定讀者個人適性化的書籍推薦服務時，將可以提供非常有用的參考資訊

This paper uses the records of readers’ borrowing history as the source data of exploration. The specific borrowing history of an arbitrarily selected reader is aimed at as the subject of exploration, to which a set of association rules is applied to find adaptive book recommendations for the reader from two aspects. From the aspect of applying association rules and that of applying association rules with weighted interest-items, antecedents of these two aspects are respectively contained in the records of the reader’s borrowing history. The tendency revealed in the two aspects through the characteristics of the reader’s reading history may provide the direction and principle for adaptive book recommendations. According to this methodology of exploration, a system for finding adaptive book recommendations is designed and constructed. The results of this exploration and construction will be particularly useful in the library service of book recommendations.