| Summary: | 碩士 === 元智大學 === 資訊工程學系 === 107 === Traditional large-scale forums have been popular in Taiwan such as PTT, Mobile01, and Bahamut. Many netizens will carry out various forms of information exchange on them. There is often a long discussion thread, which contains huge information and opinion leaders with various wind directions. People will believe certain arguments and lead the wind and influence more because of the behavior of these opinion leaders. For the general public and professional community analysts, it is difficult to find out the key messages and find the leader of the information when browsing this type of large forum. An effective and easy-to-use visual tool is necessary in order to quickly browse a large number of articles, finding hot topics and find opinion leaders who are influential in the forum. We propose a visualization system for a specific thread in the forum analyzing the content of the articles and the attributes of the author, to help users observe the overall status of the discussion thread, the feedback of the article, and the interaction between the article discussers and the dynamic influence of the author.
We split the visualizing discussion thread problem into three levels: the macroscopic level, the goal is an overview of the entire discussion thread; the mesoscopic level, observing the author's attributes and the dynamic hot topics; the microscopic level, the discussion thread in the article, including the attributes of all respondents and the group relationship between respondents. we propose solutions to the three perspectives and design corresponding visualizations. We also selected a number of features related to their role and show their influence. We implement interactive systems to help users discover the information they want to gradually explore to assist users and increase system operation efficiency.
At the end of the paper, we conducted case studies on Taiwan's most popular forum, PTT, to demonstrate the availability of the system.
|
|---|
