Application of User Clustering Based on Topic Modeling

• Main Authors: Tzu-Ching Lin, 林子敬
• Other Authors: 陳景祥
• Format: Others
• Language: zh-TW
• Published: 2018
• Online Access: http://ndltd.ncl.edu.tw/handle/kc222t

碩士 === 淡江大學 === 統計學系應用統計學碩士班 === 106 === With the advancement of network technology, social media has been widely used by the public. People express their opinions on social networks such as facebook or twitter. These remarks can reflect a lot of information about users, such as favorite things, ideas or tendencies. We can use these information to group users for facilitating subsequent research analysis or gaining business benefits. In this article, we collect the documents sent by users in the social network and using the topic model to find out which topics commonly used by each user. After finding the topic distribution for each user, we can cluster them by using some clustering analysis methods such as k- means, affinity propagation, etc. We also consider the time effect and explore the changes in the user''s topic and clustering in each time slice. Finally, We also uses the PTT data, showing the effect of the user clustering and some discovery under the Chinese documents.