基于手机短信信息流的热点事件识别  

HOT EVENTS RECOGNITION BASED ON MOBILE PHONE SHORT MESSAGE INFORMATION FLOW

作　　者：刘金岭[1] 王新功[2] 周泓[1] 

机构地区：[1]淮阴工学院计算机工程学院,江苏淮安223003 [2]沧州师范学院计算机系,河北沧州061001

出　　处：《计算机应用与软件》2012年第10期200-204,共5页Computer Applications and Software

基　　金：河北省科技支撑计划项目(10213581)

摘　　要：随着手机通信技术的快速发展,短信文本数据流作为载体携带了越来越多的信息资源。为了挖掘出短信文本信息流中富含的一些舆论焦点、社会热点问题,提出热点事件识别算法。该算法采用词共现频度定义特征词相关度,综合前导信息集合及信息产生频率给出短信文本相似度,进而设计基于Single-Pass聚类模型的短信文本信息流在线分检算法。在真实数据集上实验表明,各项性能指标都有不同程度的提高。With the rapid development of mobile phone communication technology, the short message text data stream as a carrier brings increasing information resources. In order to dig out the focus of public opinions and hot social events contained in short message information flow, we propose a hot events recognition algorithm. First, the algorithm adopts word co-occurrence frequency to define the correlation degree of feature words. Then through integrating the preamble information collection and the generation frequency of information it gives the short message＇ s similarity. Furthermore we design an online sorting algorithm for SMS text message stream based on single-pass clustering model. Experiments based on real data set indicate that its various performance indicators have been improved to some extent.

关 键 词：短信文本 信息流 热点事件 Single-Pass 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]