一种新型的网络社区高影响力主题提取方法  

A NEW EXTRACT METHOD OF HIGH-IMPACT THEME BASED ON NETWORK COMMUNITY

作　　者：吴亚男[1] 杨云[1] 

机构地区：[1]陕西科技大学电气与信息工程学院,陕西西安710021

出　　处：《陕西科技大学学报（自然科学版）》2011年第1期138-141,共4页Journal of Shaanxi University of Science & Technology

摘　　要：针对网络社区传统热点话题选择方法中存在的不足,如简单数字统计、没有考虑主题内容等,提出了一种基于社区主题内容的高影响力主题提取方法.运用传统的词语权重计算方法TF-IDF并结合网络社区的特点来定义词语权重,再根据网络社区信息传递的特点得到词语的影响力并计算出词语间的关联度,然后深度挖掘潜在关键词并最终构造出由若干个完全图组成的无向图G,从而得出高影响力主题类型.该方法能够准确提取出当前社区的热点主题并能够在一定程度上对近期热点信息进行预测和判断.The traditional method of extracting high-impact theme based on network community which does not consider the content of theme and just simply statistic.To change the traditional weaknesses,the paper proposes a new method based on thematic content for extracting high-impact theme.The system descript the weight of words by using the TF-IDF and the features of network community,then gain the influence of word in community according to the feature of information transmission and calculate the relational degree between words.In addition,mining the potential key word deeply to create an undirected graph G consisting of several complete graphs.As a result,achieve the extraction of types of high-impact themes.This system can accurately extract the current hot topics and predict and judge recent focus information.

关 键 词：权重 词语关联度 潜在关键词 

分 类 号：TP393.01[自动化与计算机技术—计算机应用技术]