多社群協同式過濾之文件推薦研究

Abstract

知識是組織中重要的智慧資產，透過良好的知識分享和知識管理的機制，智慧資產才夠有效提昇組織價值和能力。近年來，文件推薦方法已成功應用於知識管理和知識分享平台的開發，透過使用者相同興趣的分析，預測使用者對知識的喜好程度，並提供使用者感興的知識文件，以達到組織內知識文件的分享，加速智慧資產的累積。本研究主要利用多社群的概念，結合協同過濾推薦方法達到組織知識分享。研究分成兩大部份，第一部份是多社群分析，考量使用者多元的興趣，以自動化的方式，建立組織中不同知識領域的知識社群，故組織中的每個使用者可以參與一個以上的知識社群。第二部份是結合多社群與協同過濾之文件推薦，我們提出四個以多社群為基礎之文件推薦方法，分別為最大預測值方法(MPM)、主要社群方法(PCM)、使用者權重方法(UWCM)以及群內相似度方法(SCM)。實驗結果顯示，以多社群為基礎之文件推薦方法皆優於傳統之協同過濾文件推薦方法，其中最大預測值方法(MPM)最能夠有效提昇文件推薦之準確性。

Knowledge is an important asset to improve the value and capability of organizations. How to use these intellectual assets is a critical issue. Some researches indicate that a well knowledge sharing or knowledge management techniques can increase circulation of intellectual assets effectively. Document recommendation methods and community techniques have been applied to establish knowledge sharing successfully. Our research has two stages to implement a knowledge sharing platform. First stage is to create communities automatically. A community represents a knowledge interest domain. Generally, a user can join more than one community to form a multi-community environment. The second stage is to provide document recommendation based on multi-communities. We propose document recommendation methods based on multi-communities, including Max Prediction Method (MPM), Primary Community Method (PCM), User Weight in Community Method (UWCM) and Similarity in Community Method (SCM). The experimental result shows that document recommendation methods based on multi-communities outperform traditional method such as CF, and MPM is the best method to improve the quality of document recommendation.