Screening Analysis And Application Of Film Review Based On Random Forest

Recommendation analysis on Internet movie reviews is a new way to promote the development of the movie industry.The rapid progress of movie review sites has also played an active role in promoting it.Effective movie reviews can help the audience understand its pertinent information about the movie more quickly and accurately.However,the application of Internet movie review information has challenges such as difficulty in screening and unclear classification.For most users,it is sometimes difficult to choose worthwhile movie reviews as a reference.In this thesis,machine learning algorithms is used to establish a two-category model based on the movie review data to identify operational movie reviews for users as a reference,which has practical application in a certain extent.We focus on the analysis and application of the screening problem on the analysis of movie review data.The LDA(latent Dirichlet allocation)model and method is used to extract the feature words related to the evaluation of the real content of the movie from the movie review data,and the semantic similarity of the feature words of the movie is calculated by combining the word2 vec model.The aim is to expand the set of feature words of the movie.Combined with the features on Douban website and the classification of common film basic styles,we select the film comment features from three aspects: the content of film comment text,film comment metadata and film feature words.Following contents are related to our selected film comment features,such as text length,sentence number,average sentence length,text similarity,score difference,comment time limit,comment tendency,number of film feature words and sentence average.The number of feature words and the weight of feature words are discussed.The movie review data obtained from Douban website is used as the experimental data for our application analysis.The random forest method is used to construct the movie review model for classification.By using different feature combinations,we build a random forest model,which is based on the film comment classification model.Combined with the comparison on accuracy,precision,recall and f-value,the validity of the movie comment feature is verified.Furthermore,the influence of movie comment feature on the classification model of movie comment is further analyzed.At the same time,the film comment classification model based on random forest is compared with other classification models to verify the effectiveness of the model in Film Comment classification.