An effective community detection based approach to solve the cold start problem in recommender systems

Abstract

With an increasing market size, electronic commerce is a driving force newlinefor a business to enable a firm or individual for online shopping or marketing over newlinean electronic network, typically the internet. Because it reduces the transaction newlinecost, low energy cost and provides access to the global market. At the same time, newlinethe explosive growth of the Internet has increased the volume of information. newlineThus, e-commerce applications to confront the information overload problem in newlinewhich users are finding the right information at the right time is difficult. newlineFrom the business point of view, recommender systems have thepotential to solve the information overload problem. Recommender system task is to recommend items and also help the user in selecting/purchasing items from an overwhelming set of choices.Collaborative filtering recommender system is one of the most successful approaches in an e-commerce website. Collaborative filtering is a newlinemethod that provides personalized recommendations, based on preferences newlineexpressed by a set of users and calculates the similarity between customer newlinepreference ratings to identify like-minded customers and predict their product newlinepreferences.Although the collaborative filtering recommender system successful, it newlinesuffers from major issues including cold-start and scalability. The adequate or newlinesufficient information is not available for a new item or user, the recommender newlinesystem runs into the item/user cold-start problem. The new item cold-start problem newlineoccurs when the item is new and it does not have any ratings. The new item cannot newlinebe recommended unless a user has rated it before newline newline