Author : Chitresh Kumar Singh 1
Date of Publication :22nd May 2018
Abstract: Recommender systems are used to provide personalized recommendations to users in the e-commerce industry. Two main approaches for the recommender systems are collaborative filtering and content based filtering. In collaborative filtering, a user’s preference is calculated by his similarity to the other users. If a user has already rated or bought an item, then the preference for another user is calculated by his similarity to the other user. In content based filtering, the approach is item based, which means that if user has already rated or bought an item, then his preference for another item is based on the similarity of the first item to the second. Both of these filterings are combined in the form of hybrid recommender systems, and when weights are assigned to these recommendations, the system so developed is known as aweighted hybrid recommender system. An often neglected feature in recommender systems is that of ‘Serendipity’. Serendipity means introduction of newer items into the recommender system, which are likely to interest the user. In this paper we have presented a suitable model, based on the feature combination technique, which introduces serendipity feature into the recommender systems.
- G. Antolic, L. Brkic, “Recommender System Based on the Analysis of Publicly Available Data”, IEEExplore, MIPRO, May 2017.
- M.Shvartz, M. Lobour, Y. Stekh, “Some Trends in Modern Recommender Systems”, MEMSTECH 2017.
- Mustika A., Rahmad M., Indra B., “Implementation of Weighted Parallel Hybrid Recommender Systems for e-Commerce in Indonesia”, IEEExplore, ICACSIS 2016.
- Cai-Nicolas Ziegler, “Book-Crossing Dataset”, http://www2.informatik.unifreiburg.de/~cziegler/BX/,Augus t /September 2004.
- Toby Segaran, “Programming Collective Intelligence”, J. O‟Reilly Media, Inc., Aug.2007, ISBN-10:0-596-52932-5