The Study And Application Of Public Sentiment Analysis Based On Text Mining Technology

Compared with traditional data mining,the task of text mining is discovering potential and comprehensible knowledge from unstructured/semi-structured data.Text Mining involves Natural Language Processing,Machine Learning and Computer Linguistics and so on,and has been widely used in various fields.Public sentiment is composed of a mass of Internet data which is a kind of typical semi-structured data,so it is significant to track public sentiment directions by means of text mining technology.The objective of this paper is aiming to applying text mining technology to the fields of public sentiment analysis based on in-depth research.This paper gives an introduction of research at home and abroad concerning text mining and public sentiment analysis firstly.And subsequently,the author introduces algorithms of k-means and single-pass to realize hot topic recognition function,and try to improve algorithm performance from 3 aspects.The essence of warning and monitoring public sentiment is the automatic classification for internet data.The author chooses KNN to realize sensitive public sentiment classification.The K-nearest neighbor will probably exert harmful influence on performance on account of diversity of sample size and density,and it is the reason why the Fixed Weighing Factor is raised to weaken the negative impact.In this paper,the author calculates the semantic similarity between each word with Hownet in the first place,and then sentence,paragraph and text successively by the models of Similarity and the Maximum Sequence as account of the complexity of Chinese semantics in Emotional Tendency Analysis.In conclusion,this paper realizes public sentiment analysis from aspects of Hot Topic Recognition,Sensitive Public Sentiment Warning and Monitoring,and Emotional Tendency Analysis by the means of text mining algorithm.In the end,the author summarizes the research work and makes a plan for future work.