Abstract— Entertainment industry want to increasing performanceof box office it is related revenues important for movies, when estimatedorganization decided budget of movie its dependent upon a scripts. In thispaper we propose the prediction of box office performance using movie scripts.
Inthis paper proposed extract three levels of textual features 1) Genre andcontent 2) semantics 3) bag of words, Domain knowledge of screen writing forthe scripts processing technique of the input and natural language, Thesetextual features and variables define a distance metric across scripts. Thereare used as a inputs for kernel-based approach to review box officeperformance. our proposed method prediction box office performance collectionis more accurately i.e 29 percent lower mean squared error (MSE)) compared tobenchmark methods.
Keywords: Entertainment industry, text mining, kernelapproach, semantics.