| Summary: | 碩士 === 國立臺灣大學 === 資訊管理學研究所 === 106 === More and more people share their life and photos on online photo sharing social networks. Therefore, in this thesis, we propose a framework to mine user behavior from user-generated contents on a photo sharing social network to help better understand user behavior and discover how user behavior changes over time. The proposed framework contains four phases. First, we extract four user’s attributes from user-generated contents for each user namely, activeness, recognition, user topics and geographical topics. Second, we employ the fuzzy c-means clustering method to respectively cluster together similar activeness attributes and similar recognition attributes. Third, for each user, we convert the user topics into an UT-set and apply the Apriori method to mine frequent UT-patterns. Similarly, we convert the geographical topics of each user into a GT-set and apply the Apriori method to mine frequent GT-patterns. Finally, we analyze user behavior by investigating the relationships among clusters formed and frequent patterns mined. The experiment results show that the proposed framework can identify user behavior for various kinds of users, and provide valuable managerial insights for managerial applications.
|
|---|
