Prediction of Film Score: Based on Character Relations
Download as PDF
Film industry plays an important role in people's leisure life, and it has been booming even more important in recent years. With increasingly amount of film and television practitioners joining the industry, the competition in the film industry is becoming progressively fierce. Therefore, the significance of writing good movie scripts has appeared in the field. In this article, we constructed the character relationship network with soft¬ware python and extracted some information from the network and constructed a linear model with least square method based mainly on 45 randomly selected films to help the relevant practitioners build a better character relationship network in the movie script. The appropriate linear model M2 is selected by using the 5-fold cross-validation method. Then we built the random forest model MR and also used 5-fold cross-validation method to test the model. The results show that the model MR has a much better predictive effect on movie scoring compared with model M2. The model reveals that some parameters of the character relationship network in the script are able to determine the final score of the film to some degree, which will provide a reference value for the creation of the script in the film production.
Linear model; random forest; relationship network of the characters; 5-fold cross validation