基于主题的文本可视分析研究  

A Survey on Topic-Based Visual Text Analytics

作　　者：巫英才 崔为炜 宋阳秋[1] 陈杨[1] 刘世霞[1] 

机构地区：[1]微软亚洲研究院,北京100190

出　　处：《计算机辅助设计与图形学学报》2012年第10期1266-1272,共7页Journal of Computer-Aided Design & Computer Graphics

摘　　要：文本可视分析是一个跨学科的领域,涉及文本数据挖掘、计算机图形图像以及人机交互等各方面的知识和技术,可以帮助用户以可视分析的手段交互地分析海量的文本数据内容,提供及时的反馈,发现异常和规则,提取知识以及获取洞察.已被应用在国土安全、商业智能分析以及金融分析等很多不同的领域,受到国内外学术界、工业界以及政府部门越来越多的重视.文中首先简要地介绍了文本可视分析的一般流程;然后系统地介绍了典型的文本分析和可视化技术,并着重讨论这2类研究的最新技术以及发展;最后对全文进行了总结并展望了文本可视分析面临的四大研究挑战:海量数据规模、复杂数据的不确定性、数据融合以及人机交互.Visual text analytics is a multi-discipline research area involving text mining, computer graphics and imaging, and human computer interaction. It helps users analyze a large amount of text documents interactively and visually to provide timely feedback, find anomalies, detect patterns, and gain insight. It has been used in many fields such as homeland security, business intelligence, and financial analysis and has received much attention from academia, industry, and government. This paper introduces a general pipeline of visual text analytics, followed by a survey of existing text mining and visualization techniques. We finally suggest a few potential research challenges, namely, the challenges of scalability, uncertainty, heterogeneity, and user interactions.

关 键 词：文本信息可视化 文本数据挖掘 信息可视化 可视分析 人机交互 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]