A Study Of Irony Recognition In Social Network Reviews

With the rapid development of information technology,the online document becomes the main information carrier,and source of information in daily life.With the advent of the Internet Web2.0 era,people would publish,share,and disseminate information on social media,rather than being constrained to accept information from the portal.With the participation of users in the process of information generation,the content of network information has become more and more diverse.These views on the content of public opinion analysis,e-commerce and other aspects have important significance and practical value.A research hotspot of these text views is sentiment classification,in which often accompanied by irony who used to express subjective and deep-rooted views.While the use of irony will increase the difficulty of sentiment analysis,before improving the accuracy of sentiment classification,we need to explore irony recognition,and that is what we studied in this paper.This paper studies irony recognition from two aspects: one is the rule-based irony recognition method and another is the irony recognition method based on machine learning.For the rule-based ironic recognition method,this paper proposes two innovative rules: break rules and violation of common sense rules.Contradictory relationship is introduced to the violation of common sense rule.By using antonyms rules and Negative word rules of contradictory relationship to determine whether the text is a violation of common sense rules.A text who meets one of the above two rules will be classified as irony.For the text does not meet the rules,machine learning methods will be used to train a classifier to recognize the irony.In the process of training classifier,the feature system is constructed: The English word,the specific modal word,the specific vocabulary in the text,the network vocabulary,the homophonic word,and the continuous punctuation mark.All the words are also incorporated into the feature system.When in contrast to irony recognition with only machine learning method,it is found that irony recognition that proposed in this paper,which combined rule-based method and machine learning method together is more efficient.