Algorithms for book categorization

Abstract

The amount of digital books or ebooks is increasing day by day. This large data needs to be stored in proper manner. Book Categorization is a task of assigning category or set of appropriate genres to a book. The goal of this thesis is to classify books with relevant genres. There are many approaches that had been developed for text categorization like Sup port Vector Machine(SVM), Neural Text Categorizer(NTC) etc. We implemented existing algorithms with di erent datasets. We also implemented existing feature selection methods to select features. We proposed new approach on the basis of existing algorithms. In our approach we classi ed books using its properties like title, description, genre, author and did text mining on it. In the learning part, we created dictionary of keywords from books description and title, then assigned genres to the keywords. In the classi cation part, we assigned genres to a book. For classifying the books, we extract dataset from web pages using web scrapping. We also computed accuracy and training time of all implemented algorithms. Com pare the results of all existing algorithms with new proposed classi er. In proposed approach we are trying to reduce time of training when large amount of data is considered