利用靜態、動態和互動視覺化技術去呈現網上評價

Abstract

With the progress made in the Web 2.0 technology, people are more likely to share their opinions, status and feelings via social media. But due to the explosion of data in volume on the Internet, people are being overwhelmed by the overloading information and begin to get lost and confused. Since people are used to find and read the online reviews before making their further decisions recently, a tool which is able to summarize the opinion data and reveal the explicit social image of a particular target should be proposed and introduced. According to the previous work, Ting-Ying Huang has proposed a visual representation which is able to display the overall summarization of targets’ reputations. However, in most situations, people are likely to be interested in a specific aspect of individuals. Therefore, some additional visual approaches should be proposed to gain the meaningful clues. In this thesis, we aim at refining both the reputation evaluation methods and the visual representations which are suggested by Ting-Ying Huang. To accurately estimate the explicit social images of several particular targets, we analyze the opinion data at sentence level by incorporating the knowledge of Natural Language Processing. Additionally, we propose several static visualizations to reveal the detailed information such as the temporal evolutions, topical summarizations, content diversity among various features and social mediums. Moreover, to enhance the graphical perception and comprehension, we create an animated bubble chart which is able to exhibit the trends and changing patterns over time. Furthermore, we offer a fully interactive mean for users to examine and explore the opinion data via any viewing aspects.