Construction And Verification Of Chinese Cinema Box Office Prediction Model Based On Text Mining

The film industry is an important part of China’s cultural industry,whose industrial environment has been continuously improved.The flourishing development of the film industry has demonstrated China’s cultural self-confidence and made great contributions to China’s cultural soft power.From 2003 to 2017,China’s movie box office achieved a rapid growth of 900 million to 56 billion.The rapid development of the film industry has attracted all types of investors to enter,meanwhile,the film industry is also one of the high-risk industries,with high incomes as well as accompanied with high risks.Due to the high risk and uncertainty of the box office,it is necessary to identify and analyze the factors affecting the box office of the movie,and predict the film box office so as to guide the film investors and relevant stakeholders to make investment in the films.At present,regarding the factors affecting the box office of the movie,most of the domestic researches stay in the study of the internal factors of the film itself,where the data dimension is relatively simple.With the development of the Internet,the influence of Internet word-of-mouth on the sales of goods is becoming more and more significant.As a special one-time consumer goods,movies also need to incorporate the Internet word-of-mouth factors in external factors into the influence factors of the movie box office.In addition,the life cycle of the film is usually more than two weeks,and the box office is gradually increasing as time grows,and will be affected by user comments during the its release.However,the current forecast for the box office of the movie is to directly and finally predict the final box office,and there will be large difference between the result of direct prediction and the actual box office.This paper analyzes the research theories related to Internet word-of-mouth and movie box office,and combines the content obtained from text mining of movie reviews to identify the influencing factors of the movie box office.The web crawler was used to collect the research data,and the text mining technology was used to excavate the collected comments in cat’s eye app.The prediction model of multivariate linear regression and BP neural network for the first-day movie box office and the final movie box office was established by using the films released in China in 2016 and 2017 and combining the identified factors influencing the movie box offices,and the results of the model are analyzed and tested for stability.Finally,the films released in 2018 verified the effect of this model.The experimental results show that the multivariate regression model and the BP neural network model have a good fit to the first day box office and the final box office.When the film was released in 2018,for the first day movie box office forecast,the error rate of the multivariate regression model and the BP neural network model is 29.98% and 25.58% after removing the films whose first-day box office is less than 5 million yuan.For the final movie box office forecast,the multivariate regression model and the BP neural network model are both around 20%.The prediction error rate of the former box office of the film is concentrated in the range of 20%-25%.The effect of prediction model established in this study is similar to that of former studies.In addition,more verification samples are used in this paper,which can better explain that the prediction effect of the final prediction model of film box office is more stable.Generally speaking,the prediction effect of BP neural network model is better than that of multiple regression model.The model results have certain reference value for the film investors and the cinema lines.Finally,based on the results of empirical analysis,this paper proposes countermeasures and suggestions for improving the box office of the movie from the aspects of film production and marketing.The opinions put forward from the film production are enriching the film type,innovating the film technology effect,and enhancing the film’s main lineup.The five points of improving the level of script creation and creating a series of branded films,the recommendations from the marketing side are to choose a reasonable screening period,improve the film network score and improve the film network word of mouth.