@INPROCEEDINGS{KarydisNGS2009,
  AUTHOR =       {Ioannis Karydis and Alexandros Nanopoulos and Hans-Henning Gabriel and Myra Spiliopoulou},
  TITLE =        {Tag-Aware Spectral Clustering of Music Items},
  BOOKTITLE =    {International Society for Music Information Retrieval},
  YEAR =         {2009},
  pages =        {159--164},
  abstract =     {Social tagging is an increasingly popular phenomenon with substantial impact on Music Information Retrieval (MIR). Tags express the personal perspectives of the user on the music items (such as songs, artists, or albums) they tagged. These personal perspectives should be taken into account in MIR tasks that assess the similarity between music items. In this paper, we propose an novel approach for clustering music items represented in social tagging systems. Its characteristic is that it determines similarity between items by preserving the 3-way relationships among the inherent dimensions of the data, i.e., users, items, and tags. Conversely to existing approaches that use reductions to 2- way relationships (between items-users or items-tags), this characteristic allows the proposed algorithm to consider the personal perspectives of tags and to improve the clustering quality. Due to the complexity of social tagging data, we focus on spectral clustering that has been proven effective in addressing complex data. However, existing spectral clustering algorithms work with 2-way relationships. To overcome this problem, we develop a novel data-modeling scheme and a tag-aware spectral clustering procedure that uses tensors (high-dimensional arrays) to store the multigraph structures that capture the personalised aspects of similarity. Experimental results with data from Last.fm indicate the superiority of the proposed method in terms of clustering quality over conventional spectral clustering approaches that consider only 2-way relationships.},
}