Topic Extraction And Analysis Of Online Public Opinion

Topic automatic extraction and analysis is the key technology of online public opinion early-warning and analysis, and has become a hot spot in the area of data. The technology collects online public opinion data from social network and media website, detects topics, and mines the topics with multiple points of view and deep degree. This paper studies how to extract topic of online public opinion fast and accurately and analyse it with public opinion index.The technology including data collection and pretreatment method, named entity recognition technology, topic detection and topic index analysis technology of online public opinion data. Some major contributions are listed as follows:(1)We optimize the data acquisition mode of online public opinionand propose a methodof multi-level lable and multi-target keyword extraction. On the one hand, the algorithm canfiltrate noise data such as advertisement and avoid repeat acquisition. On the other hand,because of complication of data structure it can extract and preprocess data such as wordsegmentation, title mark, paragraph mark, website mark, website type mark, keywordsextraction.(2)Based on statistics and rules, a method for extracting names of online public opiniondata is proposed. We analyse geographic names of online public opinion data by establishingregional correlation table, mainly to analyse places with the same name and attribution ofplaces.(3)We propose a method based on sample hierarchical clustering method for extractingtopic of online public opnion.The data of online public opinion has some featrues such aslarge storage,complex structure, wide spread. In order to optimize the algorithm performance,at first,we reduce the number of articles by sampling data of online public opinion,thencluster for extracting topic.(4)Comprehensive analysis of public opinion topics which are exacted with three levelindex of public opinion. It can analyse the influence of public opinion all sides.