Hybrid Genre Recognition Based on Movie Script Features

摘要

Identifying a template and categorizing the movies with a Hybrid genre is an illustrious issue with many tries and approaches to resolve it. A Hybrid genre is a genre that blends themes and elements from two or more different genres. From the parallel analysis of various movie scriptswith Multiple subgenres, 9 recurrent features namely the focus of concentration of the main character in the movie, changes of locations, proportions of verbal and non-verbal communication in the movie, number of scene changes, number of night/day scenes and number of Interior/exterior scenesare identified and variations in the proportions of these features will give the clarity to classify and identify the movies with hybrid genre. This work is a step towards automatic identification of movie script with hybrid genres given a random movie script. The methodology for automaticsubgenre identification of movies uses 9 script based features and the correctness of the proposed classification is verified with Naive Bayes machine learning techniques. Thus an accuracy of 85% is achieved in identifying the 5 most popular hybrid genres namely Action comedy, Horror comedy,Romantic comedy, Sci-fic comedy and Thriller comedy and also the experimental results shows that the proposed work performs better than the existing techniques.