IRuSL: Image Recommendation Using Semantic Link

Abstract

Popularity of social networking sites and introduction of e-commerce has led to an increase in number of multimedia resources generated every year. The growth in the number of images that are being uploaded has brought a need to develop systems that can store, process and organize them as how and when needed. A survey conducted on working ability of famous e-commerce systems has brought into light the need for semantic integration in recommendation systems. This paper proposes a framework in which images are recommended to users based on their semantic integration. This integration is made possible by using tags and surrounding text of an image to determine their semantic association. In this paper four semantic relatedness methods are used. The methods are Jaccard similarity of sets, Dice, edit distance and Overlap, these techniques are compared to find the best semantic recommendation. The proposed system has wide application in recommendation systems such as image recommendation in social networking, books recommendation, movie recommendation and product recommendation in e-commerce websites.