Book Recommendation using Cosine Similarity

Rofeca Giri Rymmai, Saleema JS

Abstract


Recommendations have been a driving force for the sale of products. For years, recommendations have always been based upon the product review of customer which in a later stage, was upgraded to personalization, whereby customers having similar purchasing patterns are clubbed together and their preferences are interchanged. This paper takes on a different approach when it comes to recommendation of books. Books are usually referred based on the author, genre and book ratings. Each book has its own plot summary which is different from the description that gives only a gist of the book. These plot summaries can be used to find how two books are very much alike. The paper explains how this is achieved by the calculation of the degree of likeness between the two plots based on the terms that are used. It involved the application of text mining to find the significant terms that will help to contribute to finding the angle of closeness using the mathematical calculation of cosine similarity.

Keywords: Recommendation system, Information retrieval, TF-IDF score, Cosine similarity.

Full Text:

PDF


DOI: https://doi.org/10.26483/ijarcs.v8i3.2995

Refbacks

  • There are currently no refbacks.




Copyright (c) 2017 International Journal of Advanced Research in Computer Science