Research On Movie Box Office Forecast Based On Sentiment Analysis Of User Reviews

As an important field of entertainment service industry,movies have enriched people's lives and provided people with high-quality spiritual food,while gradually becoming an important part of China's economic market.However,China's film industry started late and developed slowly,and the domestic market's marketing mechanism is not perfect.Predicting the movie box office can help the theater layout in advance and increase box office revenue.Therefore,this paper studies how to use movie reviews to predict the movie box office.The main contents of this article are as follows:(1)This article introduces the development of movie box office prediction at home and abroad from the two aspects of the research of movie box office impact factors and the research of movie box office prediction models,and introduces the mainstream movie box office prediction methods.(2)In view of the problem that the CHI algorithm may produce extraneous features,this paper proposes a CHI-SVD feature extraction algorithm.This algorithm uses the SVD algorithm to perform dimensionality reduction on basis of the features extracted by CHI,so that it can retain the basis of the original features.The dimensionality reduction process can effectively reduce the time cost.It is found through experiments that the performance of CHI-SVD algorithm is better than that of CHI algorithm.(3)In order to improve the quality of sentiment dictionaries and to better classify sentiment,this article combines sentiment classification methods based on sentiment dictionaries and statistical information to reduce the disadvantages that sentiment classification will be affected by the construction of sentiment dictionaries,thus making sentiment The effect of classification is better.(4)In order to further accurately predict the box office of the movie,this paper proposes a random forest regression algorithm that combines SVR and multiple linear regression algorithms,and combines the emotional characteristics and the audience's rating of the movie into the random forest algorithm proposed in this paper.In addition,this paper also combines the ranking of movies compared to other movie users of the same genre to replace the genre features that are difficult to quantify,and introduces it into the random forest algorithm proposed in this article.The effectiveness of the random forest regression algorithm is verified through experiments.