PerSentiment: A personalized sentiment classification system for microblog users

• Main Authors: SONG, Kaisong, CHEN, Ling, GAO, Wei, FENG, Shi, WANG, Daling, ZHANG, Chengqi
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2016
• Subjects: Databases and Information Systems
• Online Access: https://ink.library.smu.edu.sg/sis_research/4571; https://ink.library.smu.edu.sg/context/sis_research/article/5574/viewcontent/p255_song.pdf
• Institution: Singapore Management University

Microblogging services are playing increasingly important roles in our daily life today. It is useful for microblog users to instantly understand the sentiment of a large number of microblogs posted by their friends and make appropriate response. Despite considerable progress on microblog sentiment classification, most of the existing works ignore the influence of personal distinctions of different microblog users on the sentiments they convey, and none of them has provided real-world personalized sentiment classification systems. Considering personal distinctions in sentiment analysis is natural and necessary as different people have different language habits, personal characters, opinion bias and so on. In this demonstration, we present a live system based on Twitter called PerSentiment, an individuality-dependent sentiment classification system which makes the first attempt to analyze the personalized sentiment of recent tweets and retweets posted by the authenticated user and the users he/she follows. Our system consists of four steps, i.e., requesting tweets via Twitter API, preprocessing collected tweets for extracting features, building personalized sentiment classifier based on a novel and extensible Latent Factor Model (LFM) trained on emoticon-tagged tweets, and finally visualizing the sentiment of friends’ tweets to provide a guide for better sentiment understanding