| The number of teleplays that have put on record in China is increasing year by year,and more and more teleplays are broadcasting on TV.In this paper,Python crawler technology is used to get the information of 9008 TV series from douban.com,and Python also be used to process this information.According to the distribution diagram of TV series score,it is found that the rating of TV series on douban.com presents a normal distribution and moves to the right,and using the wordcloud to describe the characteristic of Chinese mained TV dramas,China Hong Kong TV dramas,China TaiWan TV dramas,the United States TV dramas,the United Kingdom TV dramas,Japan TV dramas and Kroean TV dramas.According to the variation coefficient and average score of each Chinese TV dramas director,it is found Kong Sheng,that the director of Chinese TV drams,whose TV drams score is the highest,the most stable that is also the most popular TV dramas.among the actors whose average short score of each domestic TV drama actor is more than 8000,only Hu Ge and Sun Li’s TV drama score is more than 7.0.According to the characteristics of TV dramas in different regions,this paper divides the products into three regions: China,Britain&America,Japan&Korea.Through the analysis of variance of TV dramas scoring in these three regions,it is found that there are significant differences between Chinese TV dramas scoring and British and American TV dramas scoring and Japanese and Korean TV dramas scoring.According to their TV dramas scoring,they are all worth ranking in the three regions.For: British and American TV series ranked first;Japanese and Korean TV series ranked second;Chinese TV series ranked third.This paper also analyses the variance of the main types of TV dramas in China,and finds that there is no significant difference between the scores of love TV dramas and those of costume TV dramas in China,and the scores are relatively low,which indicates that the overall quality of love TV dramas and costume TV dramas in China needs to be improved.As the number,type and production of TV series,the directors and actors of TV plays affect the public’s attention and evaluation of TV plays,this paper obtains the names,directors,actors,ratings,short reviews,types and production time of Douban TV series from Douban website.The average number of TV dramas directed by TV directors,the average number of actors participating in TV dramas,short reviews and episodes were taken as the influencing factors of TV dramas as recommended TVdramas(scoring more than 8.0 points).Logical regression analysis showed that except for the average number of TV dramas directed by TV directors,there was no significant effect on TV dramas as recommended TV dramas,the other three factors had significant effect on TV dramas as recommended TV dramas.TV plays have a significant impact on the recommendation of TV plays,and use these three factors to build a logistic regression model for the recommendation of high-praise TV plays.At the same time,the correlation analysis of each variable was carried out to find the variables related to the score,and then a linear regression model was constructed to predict the score of TV series.At the end of this paper,I used 90706 tagged comments(each comment corresponds to a tag,namely ‘recommending strongly’,‘recommendation’,‘good’,‘poor’ and ‘very poor’)to build an emotional analysis model.I compared two different models(dm model and dbow model)of doc2 vec word embedding technology to vectorize the comment.Accuracy,It is found that the accuracy of the model is greatly improved by combining dm and dbow in the model training,and the high precision model is used to predict the emotion of 30,000 unmarked comments. |
PDF Full Text Request