Content-boosted collaborative filtering approach to reduce cold start and data sparsity problems

 
 
 
  • Abstract
  • Keywords
  • References
  • PDF
  • Untitled
  • Abstract


    Recommendation systems suffer from problems related to scalability, data sparsity and cold starts, resulting in poor-quality predictions. Hybrid techniques, such as content-boosted collaborative filtering (CBCF) and/or combine collaborative filtering methods with other recommendation systems are highly essential to alleviate the drawbacks and to improve the overall prediction rate. Obviously, the combination of algorithms could make more accurate recommendations. CBCF could be used with a combination of a pure content-based predictor (pure CF), Singular value decomposition (SVD) and user-based collaborative filtering (UBCF), which improves prediction quality and thus minimizes cold start and data sparsity problems. In this paper, a modified CBCF algorithm by implicitly collecting user ratings through a user-interest model has been developed. Experimental results were tabulated.

     

     


  • Keywords


    Cold Start; Content-Boosted Collaborative Filtering; Correlation Similarity; Data Sparsity; Mean Absolute Error.

  • References


      [1] M.A. Hameed, O. Al Jadaan, and S. Ramachandram, “Collaborative Filtering Based Recommendation System: A survey, “International Journal of Computational Science and Engineering, ISSN 0975-3397, Vol.4 No.5 May 2012.

      [2] X. Su and T.M. Khoshgftaar, “A Survey of Collaborative Filtering Techniques,”Advances in Artificial Intelligence, vol. 2009, article ID 421425, https://doi.org/10.1155/2009/421425.

      [3] X. Lam, T. Vu, T. Le and A. Duong, “Addressing Cold-Start Problem in Recommendation Systems,” Proceedings of the 2nd International Conference on Ubiquitous Information Management and Communication, ACM, Suwon, Korea, (2008) January 31-February 01. https://doi.org/10.1145/1352793.1352837.

      [4] D. Sun, Z. Luo, and F. Zhang, “A Novel Approach for Collaborative Filtering to Alleviate the New Item Cold-Start Problem, “Communications and Information Technologies (ISCIT), 2011 11th International Symposium, IEEE, Cordoba, Spain, (2011) November 22-24, pp. 402-406. https://doi.org/10.1109/ISCIT.2011.6089959.

      [5] A. Sunil Kumar, M.S. Prasad Babu and B. Raja Sarath Kumar, “Implementation Hybrid System based on Content-boosted Collaborative Filtering Algorithm, “Global Journal ofComputational Intelligence Research. ISSN 2249-0000 Volume 3, Number 1 (2013), pp. 11-20.

      [6] Maddali Surendra Prasad Babu and Boddu Raja Sarath Kumar, “An Implementation of the User-based Collaborative Filtering Algorithm, “International Journal of Computer Science and Information Technologies (IJCSIT), vol.2 (3), 2011, ISSN 0975-9646, 1283-1286.

      [7] L.Zheng, Y. Wang, J.Qi, and D. Liu, “Research and Improvement of PersonalizeRecommendation Algorithm based on Collaborative Filtering, “International Journal of Computer Science and Network Security, vol.7, no.7, pp. 134-138, July 2007.

      [8] S. Xiaoyuan and M.K. Taghi, “Collaborative Filtering for Multi-class Data Using Belief Nets Algorithms,” Proceedings of the 18th IEEE International Conference on Tools with Artificial Intelligence (ICTAI'06) 0-7695-2728-0/2006, 0-7695-2728-0/06, IEEE, 2006.

      [9] Serdar Sali, “Movie Rating Prediction Using Singular Value Decomposition,” CMPS 242, Machine Learning Project Report, University of California, Santa Cruz, 2008.

      [10] MovieLens data, http://movielens.umn.edu.

      [11] Prem Melville, Raymond J. Mooney and Ramadass Nagarajan, “Content-boosted Collaborative Filtering for Improved Recommendations,” Proceedings of the Eighteenth National Conference on Artificial Intelligence (AAAI-2002), pp. 187-192, Edmonton, Canada, July 2002.


 

View

Download

HTML

Article ID: 16242
 
DOI: 10.14419/ijet.v7i4.16242




Copyright © 2012-2015 Science Publishing Corporation Inc. All rights reserved.