Sentiment Analysis Of Chinese Reviews Based On Hotel Comments

In the world,with the prosperity and flourishing of Internet technology,many network media and applications have been produced.At the same time,a great deal of Internet information has emerged,and network reviews are an important part of it.Different from the general objective documents,the subjective and emotional information contained in these opinions have very valuable reference meanings to Internet merchants,potential users and government information departments.However,if the artificial way is used to dig the huge information,it not only needs high cost and efficiency,but also is more difficult for the implementation.Therefore,the use of computer algorithm for information mining and text classification of Internet on unstructured text comments sentiment analysis technology came into being,one of which is related to the application direction of Natural Language Processing research development.In this paper,it focuses on the problem of sentiment classification in Chinese reviews.Taking the Chinese hotel comments as the experimental object,the first two chapters introduce the research and the basic technology of the problem in detail.Then,the paper will extend and improve the existing research from two aspects.The main work and results of this article are as follows:(1)This paper proposes an affective dictionary construction method based on the generalization of semantic vector space.After the integration of several authoritative dictionaries,the neural network model is added and the target of the dictionary construction is completed together with the hotel review data.The basic idea is to construct a new emotional dictionary by combining the commentary semantic vector space with the formula.In contrast with the more complete emotional dictionaries in the field,the emotion dictionary constructed by this method can better assist the emotion classification task.(2)Aiming at the merits and demerits of existing supervised learning and rule based sentiment classification,this paper proposes an emotion classification method based on semantic rule and neural network model feature fusion.This paper mainly focuses on the characteristics of Chinese hotel reviews,and applies the multilayer semantic rules to the traditional dictionary based classification method,which strengthens the accuracy of sentiment analysis of corpus text.This paper will feature orientation as a starting point,this paper proposed a feature fusion method based on sentiment classification,the regular statistical scores calculated and the process of emotional information as new emotional features into the model feature set in the template,and then use the neural network model for sentiment classification.Compared with the general classification method,this method has greatly improved in two aspects of feature representation and Chinese text sentiment classification,and experiments show that this method has better performance than the traditional sentiment classification method.