Abstract
Predicting box office revenue (BOR) of movies before releasing on big screens successfully becomes an emerging need, as it informs investment decisions on the stock market, the design of promotion strategies by advertisement companies, movie scheduling by cinemas, etc. However, the task is very challenging as it is affected by a lot of complex factors. In this paper, we first provide a strategic investigation of these influential factors. Then, we put forward a novel framework to predict a movie’s BOR by modeling these factors using big data. Specifically, the framework consists of a series of feature learning models and a prediction and ranking model. In particular, there are two models devised for learning features: (1) a novel dynamic heterogeneous network embedding model to simultaneously learn latent representations of actors, directors, and companies, capable of capturing their cooperation relationship collectively; (2) a deep neural network-based model designed to uncover high-level representations of movie quality from trailers. Based on the learned features, we train a mutually-enhanced prediction and ranking model to obtain the BOR prediction results. Finally, we apply the framework to the Chinese film market and conduct a comprehensive performance evaluation using real-world data. Experimental results demonstrate the superior performance of both extracted knowledge and the prediction results.