Title	Spatial-temporal sentiment analysis : a Sina Weibo based model and implementation
Authors	Liu, Xuanlin
Advisors	Chung, Fu Lai Korris
Degree	Master
Issue Date	2016
Absctract	Microblogging is getting more and more popular in recent years. As a kind of service that provides opinion discussing, information sharing, attitude expressing, it adapts modern people’s need for convenience and real-time. As the result that the number of microblogging users increases rapidly year by year, researchers of sentiment analysis, who study crowd emotion and attitude, start to focus more on mining from microblogging sites. Weibo, the most popular microblogging site in China, has 212 million monthly active users up to September 30, 2015, increased 48% than that in 2014. Such a huge number of users attract sentiment analysis researchers. In this dissertation, we propose a system for geological sentiment analysis on Weibo. It classifies geo-tagged posts into positive, neutral and negative. Then the results are analyzed in multi perspectives to find meaningful relations. For example, the emotion value with the area, the variation of emotion values from time or date changing, and the similarity and differences between Hong Kong and Shenzhen. The f1-score and accuracy of the final trained classifier reaches 89.04% and 89.13% respectively in cross validation stage. Using the classifier, it is found that in both cities, most of the districts that have highest emotion scores are tourism areas, and generally, emotion scores of weekends are higher than that of weekdays. Our approach and results show a way of studying from social network data, which may help computer scientists, societies, demographer, and etc.
Subjects	Online social networks -- Social aspects Public opinion -- Data processing Computational linguistics Hong Kong Polytechnic University -- Dissertations
Publisher	Hong Kong Polytechnic University
Rights	All rights reserved.