Research Of The Sentiment Analysis Technology Based On User Behavior

With the rapid development of Web 2.0 applications,mobile network and mobile teminal(e.g.4G and smartphone),SNS(Social Network Service)is becoming an important part of everyday life.The rise of SNS provides chance to sentiment analysis.It can help service providers optimize their services,help users increase experience and help government monitor public opinion.Now,for short text analyzing,many researchers use semantic lexicon or extract semantic features to analyze sentiment.And in the research of our human behavior,people can find that behavior and sentiment influence and reflect each other.So this paper start at the online users' behavior,extract the behavior features,explore aind analyze the rules between users' behaviors and their sentiment tendency.In addition,create classifier based on the users' behaviors.Behavior of SNS users has the following features:a lot of participators,normalized behaviors,easier to collect,etc.So SNS is a very good platform to research user behavior.This paper takes "Sina Weibo" for example.Start with the characters of user behavior to study user behavior's effect in sentiment analysis of SNS.Firstly,this paper systematically introduces the current work of sentiment analysis and user behavior.Secondly,according to the standard procedure of sentiment method based on character,prepare data,extract character of user behavior.Then use the statistical analysis and association mining to research how the relevance between users' behaviours and sentiment.Then build classifier based on NB and C4.5 according to practical situation.Finally,verify the result.This paper studies from network users' behavior.Through the above researches,this paper proves that there are associations and rules between SNS users' sentiment and behavior.It lays the foundation for further study.