由過去文獻分析及本研究初步統計，圖書館中許多資源未能被充份利用，而改善途徑則有賴對讀者興趣的掌握。一般而言，圖書館資訊服務所關心的問題相當明確，且圖書與讀者資料都有清楚的組織結構，因此藉由圖書借閱記錄的探勘應有助於瞭解讀者的資訊需求。本文即以世新大學圖書館所提供1996年9月23日至1998年9月22日兩年共計171,284筆的圖書借閱記錄為研究範疇，並利用關聯（Association）及分群（Clustering）等資料探勘方法，分析圖書與讀書、讀者與讀者，及圖書與圖書間的隱藏關聯與規則，如挖掘相似的圖書分類號、讀者系所，或圖書分號號與讀者系所群集。這些決策規則（Decision Rules）對瞭解並進而掌握讀者興趣有相當的參考價值。此外，研究過程中也獲得一些應用資料探勘技術的實務經驗，並深入討論其中的困難與限制，最後則對加強圖書資源利用的應用方式提出探討。

　　　　　According to various findings from our research and related studies, library resources in general are not fully utilized by users. Since understanding users' interests is an essential step to promote uses of library resources, it is useful to analyze users' interests from some unobtrusive data like transaction logs. Therefore, in this paper, we used circulation history from Shinh Hsin University Library as a basis, which included a 2-year log containing over 170 thousands circulation records, to analyze users' interests. Association and clustering methods are used to discover relationships among books and books, books and users, and users and users. The experimental results show there exist valuable relationships among circulation data for various applications, such as being a basis of recommending new books for users by referencing clusters of classification numbers. Meanwhile limitations of the circulation data, automation system, data mining techniques, and privacy issues are also discussed.