基于BTM的微博舆情热点发现

发布时间：2018-03-11 20:23

本文选题：词对主题模型　切入点：短文本　出处：《情报杂志》2016年11期 　论文类型：期刊论文

【摘要】：[目的/意义]作为一种新兴的社交新闻媒体,近年来,微博在许多热点事件的发布和传播中发挥了重要作用。但由于其文本的特殊性,传统方法不能有效地对其进行建模发现热点话题。因此,如何高效、准确地从微博数据中发现并提取有意义的热点信息是一个很有价值的研究课题。[方法/过程]提出一种基于BTM模型的微博舆情热点发现方法。首先,对微博文本采用BTM建模,改进TF-IDF权重计算算法,以适应微博短文本的特征。并将BTM建模结果与改进的TF-IDF权重算法结合对微博文本进行特征提取及相似性度量,然后采用K-means聚类方法发现热点话题。[结果/结论]通过对新浪微博数据集的对比实验及结果分析验证了本方法的有效性。本方法能够有效解决传统模型在文本建模中所面临的高维度和稀疏性问题,显著改善热点话题的发现质量。
[Abstract]:[purpose / significance] as a new social news medium, Weibo has played an important role in the publication and dissemination of many hot events in recent years. Traditional methods can't effectively model it to find hot topics. Therefore, how to efficiently, It is a valuable research topic to accurately find and extract meaningful hot spot information from Weibo data. [method / process] this paper presents a method for discovering hot spots of public opinion in Weibo based on BTM model. Firstly, the paper uses BTM to model the text of Weibo. The algorithm of TF-IDF weight calculation is improved to adapt to the feature of Weibo short text, and then the feature extraction and similarity measurement are carried out by combining the result of BTM modeling with the improved TF-IDF weight algorithm. Then K-means clustering method is used to find hot topics. [results / conclusion] the validity of this method is verified by comparing the data sets of Weibo of Sina and analyzing the results. This method can effectively solve the problem of traditional model building in text. High dimensional and sparse problems in modules, Significantly improve the quality of hot topic discovery.
【作者单位】：西安电子科技大学经济与管理学院;
【分类号】：TP391.1;G206

【相似文献】