Research On Mobile Brand Classification Model Based On Comment Emition Analysis

The emergence of smart phones has promoted the transformation and upgrading of society and facilitated people's lives.However,due to the asymmetry of information,ordinary users are unable to judge because of the lack of relevant information about mobile phone brand.With the development of machine learning,it becomes a reality to extract mobile phone brand features through massive information processing.Therefore,this paper proposes to use mobile phone purchase reviews by consumers,aim at classify mobile phone brand.The main work is as follows:(1)Data processing and custom dictionary construction.Through data processing and standardized methods,the statistics of high-frequency words for mobile phone reviews are completed,and a custom dictionary for mobile phone reviews is constructed,and the content of mobile phone reviews is summarized.(2)A mobile phone brand classification model is constructed.On the basis of word frequency statistics,we classify different features of mobile phone reviews,and extract features such as appearance design,product performance,photography,and so on.Then we realize clustering analysis of Related words similar to features,and use emotional dictionary to vectorize the content of reviews.Based on feature vectors,smart phone brand is labeled,and a classification model is constructed based on SVM algorithm.(3)The assessment and application scenario research of the classification model are carried out.By comparing SVM algorithm model with BP neural network,KNN classification models,verify that SVM has a higher classification effect.In addition,the scene application analysis of mobile phone brand classification model can provide decision-making suggestions for users to purchase mobile phones,which haspractical needs.