整合網路書店消費行為之圖書館圖書推薦系統

Abstract

對圖書館而言，降低資訊負荷對使用者進行資訊檢索時的苦難與提升冷門館藏的曝光度及利用率，一直是一個重要的挑戰。推薦系統是解決這些挑戰最好的工具之一，然而推薦系統卻必須面對侵犯使用者隱私與資料稀缺等問題。本研究以結合具有龐大資料量及真實消費者行為之網路書店推薦資料與圖書館館藏資料的方式當作推薦系統的資料來源，用以解決上述所遭遇的問題。借鑑協力式過濾之演算法，推演出適用於本研究之推薦度（R）公式，用以計算非透過使用者之間相似度為主的推薦方法。透過實證結果發現，透過結合網路書店資料一定程度上降低了隱私與資料稀缺問題，並提升冷門館藏的曝光度。

Information overload and enhance the usage of low borrow library collections is the challenges of Library. Recommender Systems (RSs) is one of the best tool to overcome these challenges, but its’ always come along with the issues like privacy and data sparsity. This research aim to overcome the challenges and issues above with integrating consumer behavior from online bookstore. Proposed algorithm is Reference by Collaborative Filtering to infer the R-index for the target and the recommend books. The results show that by integrating data from online bookstore can reduce privacy and sparsity issues for recommender system and enhance the visibility of the low borrow collections.