Abstract Journalists, government agency and the public may hold different opinions about the same event; these opinions flow within and across multiple social groups. Understanding opinion flow in multiple social groups is conducive to have a quick grasp of the information on social media. Our study aims to explore the opinion flow within or across the pre-defined social groups and the mutual influence of these groups’ views about the same event. First of all, to depict the opinion flow, we propose a topic transition model, and for expressing the interaction of groups’ opinion, we propose a quantitative approach. Secondly, according to 4Ws, we design a visual analytic system to let users analyze public opinion events through our system. The visual interface is tightly coupled with the analytical methods to present a concise summary of the opinion flow of events. A novel forwarding-network slicing diagram is designed to depict the forwarding process. To describe the result of the topic transition model, we also propose a topic transition graph to characterize the opinion flow. And a novel glyph is designed to visualize the characteristic of the influence. Our system provides a variety of interactions to assist users in exploring data, and we demonstrate the functionality of the system through two case studies. Graphic Abstract

中文翻译：

---

多个社会群体间意见流的可视化分析

摘要 记者、政府机构和公众对同一事件可能持有不同意见；这些意见在多个社会群体内部和之间流动。了解多个社交群体的意见流向，有利于快速掌握社交媒体上的信息。我们的研究旨在探索预先定义的社会群体内部或之间的意见流以及这些群体对同一事件的看法的相互影响。首先，为了描述意见流，我们提出了一个话题转移模型，为了表达群体意见的互动，我们提出了一个量化的方法。其次，根据4Ws，我们设计了一个可视化分析系统，让用户通过我们的系统来分析舆情事件。可视化界面与分析方法紧密结合，以呈现事件意见流的简明摘要。设计了一种新颖的转发网络切片图来描述转发过程。为了描述主题转换模型的结果，我们还提出了一个主题转换图来表征意见流。并且设计了一种新颖的字形来形象化影响的特征。我们的系统提供了多种交互来帮助用户探索数据，我们通过两个案例研究展示了系统的功能。图形摘要 并且设计了一种新颖的字形来形象化影响的特征。我们的系统提供了多种交互来帮助用户探索数据，我们通过两个案例研究展示了系统的功能。图形摘要 并且设计了一种新颖的字形来形象化影响的特征。我们的系统提供了多种交互来帮助用户探索数据，我们通过两个案例研究展示了系统的功能。图形摘要