Sentiment Analysis Of Microblog Product Reviews Based On Feature Ontology And Sentiment Lexicon

With the network gradually changes people's lifestyle,it has gradually become a platform for people to publish and analyze comments and opinions.Evaluation information can not only be used for public opinion supervision,but also can provide guidance for the improvement of the product and purchasing decision of customer.Microblogging product reviews are evaluation information of microblogging users who posted on the microblogging platform,the information is the user experience in the purchase and use of product.As the recipients of the information posted by microblogging users is generally their fans or friends,the information is often more real and pertinent which based on this reality relations,and the information also has a more direct impact on the recipients.Therefore,the study of microblogging product reviews,on the one hand can make full use of network information,on the other hand can provide reference information which is more real and effective for enterprises and users.In order to unify the synonyms and the hyponyms of the concept of product aspects,to improve the effect of aspect extraction,to improve the sentiment analysis accuracy of product aspect,this paper presents a sentiment analysis method of microblogging product reviews based on aspect ontology and sentiment lexicon.This method constructs the aspect ontology,which is used to classify the concept of product aspect and to extract product aspect.So it is conducive to calculating the correlation intensity of product aspects and sentiment words,and to calculating the sentiment values of product aspects.The TF-IDF is improved to calculate the correlation of product aspects and sentiment words.Then we find implicit product aspects through the strong correlation.We build a comprehensive sentiment lexicon for the accurate identification of sentiment information.After the product aspects-sentiment information tuples are obtained from product reviews,the sentiment values of product aspects can be calculated.Finally,the experiment on the data set of microblogging product reviews is carried out to validate the method we proposed,the experiment results show that the method is feasible and it can improve the sentiment analysis effect of product aspect.