HOSVD on Tensors and Its Extensions

  • Panagiotis SymeonidisEmail author
  • Andreas Zioupos
Part of the SpringerBriefs in Computer Science book series (BRIEFSCOMPUTER)


This chapter describes in detail tensor decomposition for recommender systems. As our running toy example, we will use a tensor with three dimensions (i.e., user–item–tag). The main factorization method that will be presented in this chapter is higher order SVD (HOSVD), which is an extended version of the Singular Value Decomposition (SVD) method. In this chapter, we will present a step-by-step implementation of HOSVD in our toy example. Then we will present how we can update HOSVD when a new user is registered in our recommender system. We will also discuss how HOSVD can be combined with other methods for leveraging the quality of recommendations. Finally, we will study limitations of HOSVD and discuss in detail the problem of non-unique tensor decomposition results and how we can deal with this problem. We also discuss other problems in tensor decomposition, e.g., actualization and scalability.


HOSVD Higher order singular value decomposition Tensor decomposition 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Burke, R.: Hybrid recommender systems: survey and experiments. User Model. User-Adap. Inter. 12(4), 331–370 (2002)CrossRefzbMATHGoogle Scholar
  2. 2.
    Chin, T.-J., Schindler, K., Suter, D.: Incremental kernel svd for face recognition with image sets. In: Proceedings of the 7th International Conference on Automatic Face and Gesture Recognition (FGR 2006), pp. 461–466. IEEE (2006)Google Scholar
  3. 3.
    Cristianini, N., Shawe-Taylor, J.: Kernel Methods for Pattern Analysis. Cambridge University Press (2004)Google Scholar
  4. 4.
    Furnas, G., Deerwester, S., Dumais, S., et al.: Information retrieval using a singular value decomposition model of latent semantic structure. In: Proceedings of ACM SIGIR Conference, pp. 465–480 (1988)Google Scholar
  5. 5.
    Kolda, T.G., Sun, J.: Scalable tensor decompositions for multi-aspect data mining. In: ICDM’08: Proceedings of the 8th IEEE International Conference on Data Mining, pp. 363–372. IEEE Computer Society, Dec 2008Google Scholar
  6. 6.
    Luo, D., Ding, C., Huang, H.: Are tensor decomposition solutions unique? on the global convergence hosvd and parafac algorithms. In: Advances in Knowledge Discovery and Data Mining, pp. 148–159. Springer (2011)Google Scholar
  7. 7.
    Nanopoulos, A., Rafailidis, D., Symeonidis, P., Manolopoulos, Y.: Musicbox: personalized music recommendation based on cubic analysis of social tags. IEEE Trans. Audio Speech Lang. Process. 18(2), 407–412 (2010)CrossRefGoogle Scholar
  8. 8.
    Nanopoulos, A., Symeonidis, P., Ruxanda, M., Manolopoulos, Y.: Ternary semantic analysis of social tags for personalized music recommendation. In: ISMIR’08: Proceedings of the 9th ISMIR Conference, New York, pp. 219–224 (2008)Google Scholar
  9. 9.
    Sarwar, B., Konstan, J., Riedl, J.: Incremental singular value decomposition algorithms for highly scalable recommender systems. In: International Conference on Computer and Information Science (2002)Google Scholar
  10. 10.
    Symeonidis, P.: ClustHOSVD: item recommendation by combining semantically enhanced tag clustering with tensor HOSVD. IEEE Syst. Man Cybern. (2015)Google Scholar
  11. 11.
    Symeonidis, P., Nanopoulos, A., Manolopoulos, Y.: A unified framework for providing recommendations in social tagging systems based on ternary semantic analysis. IEEE Trans. Knowl. Data Eng. 22(2) (2010)Google Scholar
  12. 12.
    Turney, P.: Empirical evaluation of four tensor decomposition algorithms. Technical report (NRC/ERB-1152) (2007)Google Scholar

Copyright information

© The Author(s) 2016

Authors and Affiliations

  1. 1.Faculty of Computer ScienceFree University of Bozen-BolzanoBozen-BolzanoItaly
  2. 2.ThessalonikiGreece

Personalised recommendations