The Korean Journal of Applied Statistics (응용통계연구)

The Korean Statistical Society (한국통계학회)
권호 표지
DOI QR코드

DOI QR Code

Modified Bayesian personalized ranking for non-binary implicit feedback

비이진 내재적 피드백 자료를 위한 변형된 베이지안 개인화 순위 방법

  • Kim, Dongwoo (Department of Statistics, Seoul National University) ;
  • Lee, Eun Ryung (Department of Statistics, Sungkyunkwan University)
  • Received : 2017.10.20
  • Accepted : 2017.11.08
  • Published : 2017.12.31
Download PDF
⟨ Previous Next ⟩

Abstract

Bayesian personalized ranking (BPR) is a state-of-the-art recommendation system techniques for implicit feedback data. Unfortunately, there might be a loss of information because the BPR model considers only the binary transformation of implicit feedback that is non-binary data in most cases. We propose a modified BPR method using a level of confidence based on the size or strength of implicit feedback to overcome this limitation. The proposed method is useful because it still has a structure of interpretable models for underlying personalized ranking i.e., personal pairwise preferences as in the BPR and that it is capable to reflect a numerical size or the strength of implicit feedback. We propose a computation algorithm based on stochastic gradient descent for the numerical implementation of our proposal. Furthermore, we also show the usefulness of our proposed method compared to ordinary BPR via an analysis of steam video games data.

베이지안개인화순위(Bayesian personalized ranking) 방법은 내재적 피드백 자료를 분석하는 최첨단 추천시스템 통계기법 중 하나이다. 하지만, 기존의 베이지안개인화순위 방법은 내재적 피드백 자료를 변환한 이진 자료만을 고려하기 때문에 정보의 손실이 있을 수 있다는 단점이 있다. 이를 해결하기 위해 본 논문에서는 내재적 피드백 자료의 수치적 크기에 기반한 확실함의 정도(level of confidence)를 고려하는 변형베이지안개인화순위 방법을 제안한다. 제안한 방법은 기존 방법처럼 상품간의 개인선호도에 관한 직관적인 확률모형 구조를 여전히 지니면서 내재적 피드백의 수치적 크기를 확실함의 정도로 반영할 수 있다는 점에서 유용하다. 또한 제안한 변형 베이지안개인화순위 방법을 수치적으로 구현하기 위해 확률그라디언트하강(stochastic gradient descent) 기법에 기반한 계산 알고리즘을 제시한다. 마지막으로, 스팀 비디오 게임 실제 데이터 분석을 통하여 기존방법에 비해 우수한 성능을 입증한다.

Keywords

References

  1. Hu, Y., Koren, Y., and Volinsky, C. (2008). Collaborative filtering for implicit feedback datasets. In Proceeding ICDM '08 Proceedings of the 2008 Eighth IEEE International Conference on Data Mining, 263-272.
  2. Koren, Y., Bell, R., and Volinsky, C. (2009). Matrix factorization techniques for recommender systems, Computer, 42, 30-37.
  3. Oard, D. W. and Kim, J. (1998). Implicit feedback for recommender systems. In Proceedings of the AAAI Workshop on Recommender Systems, 81-83 .
  4. Rendle, S. and Freudenthaler, C. (2014). Improving pairwise learning for item recommendation from implicit feedback. In Proceedings of the 7th ACM International Conference on Web Search and Data Mmining, 977-985.
  5. Rendle, S., Freudenthaler, C., Gantner, Z., and Schmidt-Thieme, L. (2009). BPR: Bayesian personalized ranking from implicit feedback. In Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence, 452-461.
  6. Sarwar, B., Karypis, G., Konstan, J., and Riedl, J. (2001). Item-based collaborative filtering recommendation algorithms. In Proceedings of the 10th International Conference on World Wide Web, 285-295 .
  7. Wang, S., Zhou, X., Wang, Z., and Zhang, M. (2012). Please spread: recommending tweets for retweeting with implicit feedback. In Proceedings of the 2012 Workshop on Data-Driven User Behavioral Modelling and Mining from Social Media, 19-22 .