• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

Recommender systems

Project Participants: Sergey Nikolenko, Dmitry Ignatov

 

Recommender systems analyze user interests and try to predict which items will be most interesting for a specific user at a specific moment of time. Modern recommender systems are based on collaborative filterint (Bell, Koren, 2007a; 2007b; Koren, 2008; 2009; 2010; Koren, Bell, 2011; Agarwal, Chen, Elango, 2009); they have experienced a surge of interest over the latest years based on the Netflix Prize contest devoted to developing a new movie recommendation system. Collaborative filtering methods achieve good results in recommending simple objects (books, movies) that do not contain a lot of metadata, and the recommendations are primarily based on the evaluations of numerous users. However, we would also like to use the abundant additional information which is often available, e.g., the immediate content of the recommended objects. In 2014, was planned to:

  • develop recommender systems and algorithms for the problem of recommending online radio stations based on the data from the FMHostweb service;
  • study existing developments in recommender systems, analyze recent publications on recommender systems using, in particular, automated text mining tools, including LDA model implemented in the laboratory over previous projects;
  • develop new recommender systems and algorithms with additional information on products and/or users.


Results:

* We have developed three new recommender algorithms for recommender systems with tags; experiments show that for small datasets our algorithms are much better than standard matrix decomposition approaches.

* We have prepared a detailed survey of main results and trends in the field of recommender systems based on recent publications in this field, including automated analysis with topic modeling.

Publications

V.A. Leksin, S.I. Nikolenko. Semi-Supervised Tag Extraction in a Web Recommender System. Proc. 6th International Conference on Similarity Search and Applications SISAP 2013, LNCS vol. 8199, pp. 206-212, 2013.

D.I. Ignatov, S.I. Nikolenko, T. Abaev, J. Poelmans. Improving Quality Of Service For Radio Station Hosting: An Online Recommender System Based On Information Fusion. Working papers by NRU Higher School of Economics. Series MAN ``Management'', 2014, no. 31. SSRN: http://ssrn.com/abstract=2542543, http://dx.doi.org/10.2139/ssrn.2542543.


 

Have you spotted a typo?
Highlight it, click Ctrl+Enter and send us a message. Thank you for your help!
To be used only for spelling or punctuation mistakes.