Downloads: 131

Survey Paper | Computer Science & Engineering | India | Volume 6 Issue 4, April 2017

Rating Prediction Based on Social Recommender Systems by Fusion of Collaborative Filtering Algorithms

Priyanka Vijay Chandegaonkar | K Vishal Reddy

Abstract: Recommender systems apply data discovery techniques to the matter of constructing customized recommendation or information, merchandise or services throughout a live interaction. These systems, particularly the k-nearest neighbor collaborative filtering based mostly ones, area unit achieving widespread success on the net. The tremendous growth within the quantity of accessible information and the range of tourists to visit websites in recent years poses some key challenges for recommender systems. These are manufacturing top quality recommendations, activity several recommendations per second for several users and things and achieving high coverage within the face of knowledge meagerness. In ancient collaborative filtering system the quantity of labor will increase with the amount of participants within the system. New recommender system technologies area unit required which will quickly manufacture top quality recommendations, even for terribly large-scale issues. To handle these problems we have got explored item-based collaborative filtering techniques. Item-based techniques first analyze the user-item matrix to spot relationships between completely different things, so use these relationships to indirectly figure recommendations for user. In this paper we tend to analyze completely different item-based recommendation generation algorithms. we glance into completely different techniques for computing item-item similarities (e. g. , item-item correlation vs. circular function similarities between item vectors) and completely different techniques for getting recommendations from them (e. g. , weighted add vs. regression model). Finally, we tend to by experimentation appraise our results and compare them to the essential k-nearest neighbor approach. Our experiments recommend that item-based algorithms offer dramatically higher performance then user-based algorithms, whereas at a similar time providing higher quality than the most effective offered user-based algorithms.

Keywords: Recommender Systems, Collaborative Filtering Item reputation, rating prediction, user sentiment

Edition: Volume 6 Issue 4, April 2017,

Pages: 948 - 951

Similar Articles with Keyword 'Recommender Systems'