Resolving Cold Start problem in recommendation system using demographic approach

Abstract

In today's era, the increasing demand of recommendation system has shown its applicability in a variety of applications. Such system makes use of various data mining techniques and algorithms in order to identify the accurate choice of item from billion of products for a particular user. This system is classified into various parts namely, Collaborative filtering, Content based filtering, Knowledge based filtering and Hybrid approach. However, recommendation system has inadvertently been an attack target since its inception. For instance, it is faced with many challenges such as gray sheep problems, shilling attacks, synonymy, scalability, data sparsity, cold start that affects the overall performance of the system. Among these challenges, cold start problem is of prime concern as it makes the system complex by not containing any prior rating history and involves three cases: recommendation for new user, recommendation for new product and recommendation of new product for new user. Therefore, in this paper, recommendation for new user in system has been focused. For such system, demographic trend by finding similarity between old user and new user has been followed. The proposed work is based on movie recommendation and has been implemented using MovieLens Dataset.