面向社会文本流数据探测爆发主题方法浅析

A Survey of Burst Topic Detection Towards Social Text Stream Data

社会文本流数据富含上下文环境信息、语言不规范且参与用户数量庞大。针对这类数据开展爆发主题探测需要寻找新的思路。本文对社会文本流数据的概念、特点以及爆发主题表达形式进行系统性梳理,从文本内容、时间、社会三个维度阐述探测爆发主题的主要研究思路和基本流程,分析利用社会特征(如用户参与、上下文环境、社团结构)进行爆发主题探测的主要技术方法。

Social text streams have rich contextual information and huge participants who communicate with informal steams. It needs to find suitable solutions to detect burst topics from this kind of data. In this paper, the authors comb through the concepts, the characteristics of social text stream data and the presentation forms of burst topics. It also sum- marizes the main research ideas and the basic procedures of burst topic detection towards social text stream data in three dimensions ： textual content, social, and temporal. The principal approaches to make use of social features, such as user participation, social context and community structure evolution, for burst topic detection are generally discussed.