Collaborative filtering-based recommender system

Abstract

Recommender systems have changed the way people find products, information, and services on the web. These kinds of systems study patterns of behavior to know someone’s interest will in a collection of things he has never experienced. Collaborative filtering is a popular recommendation algorithm that works to find user’s interest patterns and recommendations based on the ratings or behavior of other users or target user in the system. The assumption behind this method is to find a user with similar interest to the active user and use his/her preference for recommendation to the active user. But several issues exist in the kind of method. For example, accuracy, sparsity, and cold start. In this paper, an improved recommendation technique is proposed to address the issues identified.