Research On The Algorithm Of Latent Aspect Rating Analysis Based On Contrast Analysis In Text Mining

With the rapid development of global informatization,the extensive promotion and application of computer and network technology have directly led to the emergence of massive data in the network.In order to effectively obtain value information from massive data,data mining technology has developed rapidly.Among them,text mining,as an important branch of data mining,aims to analyze text data with rich information in the network.The thesis mainly focuses on the field of latent aspects rating analysis in text mining.On the basis of sorting and summarizing relevant historical literature,it is finded that there are two problems in the past research in this field.1.Due to the neglect of the personalized user comments,it is impossible to objectively infer the score.2.Ignoring user "subtext" messages.Contrast analysis method is applied to solve above two problems.Firstly,the user comment behavior is analyzed using the user history text data.Secondly,the user habits were compared with specific user comments by comparing and calculating the two influencing factors of users' "attention" and "emotional intensity",and then the personalized influence of user comments was removed.Meanwhile,the "subtext" information of users was found.Finally,based on the comprehensive consideration of personalized user comments and user "subtext" information,the weight of aspects was inferred and the score of aspects were calculated.In the experimental part of the thesis,crawler technology is used to obtain comment text data,and contrast analysis algorithm experiment is designed and implemented.The experimental results are verified by qualitative and quantitative methods to demonstrate the feasibility and effectiveness of the algorithm.