Research And Implementation Of Emotional Analysis For Short Texts Of Web Comments

With the rapid development of the Internet,various social network platforms such as Weibo,Post Bar,and Vibrato have become more and more important to people's lives.A large number of network users will generate a huge amount of information on various platforms and analyze such massive information.The hidden emotional polarity has great commercial and social value,so emotional analysis of the text for the social network platform has become a hot topic today.In the online comment texts,new online words emerge in an endless stream.Due to the openness of online comments,comments expressing similar semantics may change their emotional polarity in different contexts.In response to the above problems,this paper proposes a comprehensive emotional analysis system for online comment texts.Aiming at the problem of new words on the network,this paper establishes a set of new words automatic discovery and emotion recognition system.(1)In the aspect of new word discovery,the traditional statistical word-based new word discovery method is adopted.This method can automatically discover new words and save a lot of manual labeling work.In the new word emotion recognition,a new word context and emotion dictionary is proposed.The comprehensive comparison method of the distributed semantic similarity between the benchmark words and the new words is used to judge the emotional polarity of the new words,and finally the network new word sentiment dictionary is automatically established.(2)On the issue of semantic diversity of online commentary,this paper proposes an emotion classification method based on background enhancement.When emotionally classifying comments,the related news or post of the comment is used as its background,which is used as the feature of the sentiment analysis of the comment text.It is added to the sentiment classification model,which improves the ability of the model to analyze and judge emotions in different contexts.In the experiment,the background mixed feature model and the traditional sentiment classification model were compared,and the recall rate and F value were improved.