Movie Summarization Based On Subjective And Objective Features

With the rapid growth of movie production,users hardly have enough time to follow all new movies and review their favored classic movies.Therefore,movie summarization is of vital importance because it provides a brief of the movie content.For users who want to review the classic portions of their favorite movie,a summary of the movie is a good choice for it costs little time.Manually summarizing movies leads to successful summaries,but it is intellectually expensive and time-consuming.Therefore,there is a great demand for automatic movie summarization.Most previous works on automatic movie summarization rely on only movie data itself.The summary is generated based on the features extracted from the movies or the movie scripts.However,relying on only movie data can hardly grasp the audience's attentions.Since movie summarization is audience-oriented,the audience's opinions and feelings are more important for producing a qualified summary.Therefore,subjective feature based approaches are researched.Subjective feature based approaches try to understand the opinions of audience.Though this kind of features derived from user-generated data better depict the audience's opinions and excitements,they ignore the objective features of the movie itself.Furthermore,the digging of the subjective features is not sufficient.In this paper,we provide a movie summarization approach based on both the objective features and the subjective features.The objective features are from the movies,while the subjective features are from the real-time comments of the movies.Based on the movie and the corresponding comments,automatic movie summarization is performed for generating movie summary that reflects the exciting moments of the viewers,and provides the most important piece of information in the most complete manner,while eliminating possible redundancies.The contributions of this paper are:· An approach for comments-based highlights generation is proposed.In this paper,we use the features from comments to measure the attraction of movie segments and analyze the sentimental orientation of the segments,to generate highlights of movies.· Using the content of the comments when generating the summaries.In this paper,the content saliency and the content consistency of segments are measured using the comment content to generate qualified summaries.· A movie summarization approach is proposed based on the subjective and objective features.The subjective features are dug deeply from the comments of the audience,while the objective features are explored from the movies.They work together to improve the movie summaries.Our experiments carried out on movies of different genres have shown the effectiveness of our proposed approaches.