Abstract

With the growth of hierarchical data in public opinion analysis, new visualization methods that can intuitively present this kind of data are urgently needed. In this paper, we propose VineMap, a new visualization method with a vine metaphor form. Different from other public opinion visualizations, we devote more attention to visualize both the hierarchical structure of texts and the semantic orientation in content. First, we extract a hierarchical topic model from text data. Then we design a visualization based on a vine metaphor form to enable users to understand public opinion in hierarchical form. At the same time, we propose heuristic optimized strategies for the visualization layout. VineMap is applied both on unstructured text data and structured data to demonstrate its applicability. The evaluations not only show users’ perceptions to our method but also prove its good performance with respect to generation time, space utilization and visual effect.

Graphic abstract

中文翻译：

---

VineMap：一种隐喻可视化方法，用于根据文本数据进行民意层次分析

摘要

随着舆论分析中分层数据的增长，迫切需要能够直观地呈现此类数据的新可视化方法。在本文中，我们提出了VineMap，这是一种具有藤隐喻形式的新可视化方法。与其他舆论可视化不同，我们将更多的精力放在可视化文本的层次结构和内容的语义方向上。首先，我们从文本数据中提取分层主题模型。然后，我们基于葡萄树的隐喻形式设计可视化效果，以使用户能够以分层形式理解公众舆论。同时，我们提出了可视化布局的启发式优化策略。VineMap应用于非结构化文本数据和结构化数据，以证明其适用性。

图形摘要