Cluster Indexing and Collaborative Filtering Recommender System

  • K Keerthi, U Amulya

Abstract

Giving or recommending fitting content based on the nature of experience is the most significant and testing issue in recommender systems. As collaborative filtering (CF) is one of the most noticeable and well known procedures utilized for recommender systems, we propose another clustering-based CF (CBCF) strategy utilizing an incentivized/penalized user (IPU) model just with the appraisals given by users, which is therefore simple to actualize. We plan to structure such a basic clustering-based methodology with no further earlier data while improving the recommendation exactness. To be exact, the motivation behind CBCF with the IPU model is to improve recommendation execution, for example, precision, recall, and F1 score via cautiously misusing various inclinations among users. In particular, we figure a compelled improvement issue in which we plan to expand the recall (or comparably F1 score) for a given precision. To this end, users are separated into a few clusters based on the real evaluating information and Pearson correlation coefficient. A while later, we give every thing a motivating force/punishment as per the inclination propensity by users inside a similar group. Our trial results show a noteworthy presentation improvement over the pattern CF plot without clustering as far as recall or F1 score for a given precision.

Published
2020-05-27