Content-based music information retrieval using SNS tags

Abstract

Recently, increasing digital music collections make us difficult to find new favorite music which we are really interested in. To overcome this problem, content-based music information retrieval has been studied. It enables users to search similar music with users choose, however few method can improve search result interactively. In this paper, we propose a new method that allows users to reflect their search intention. By selecting SNS tags, users can naturally inform their intention to a music retrieval system, and the system utilize user feedback to modify a similarity function. we employ learning to rank technique to utilize tag information to modify a feature vector space.