ABSTRACT
With the exponential growth of data, it becomes more and more difficult to quickly obtain valuable information from massive amounts of data. Clicking and browsing of such non-scoring implicit data has also attracted more and more attention from scholars. Recommendations for implicit data generally use bayesian personalized ranking algorithm (BPR). The algorithm focuses on the difference in preferences between item pairs, and believes that users prefer items that have interacted with items that have never interacted. However, this assumption is still not specific enough for analyzing the relationship between users' items. Therefore, this paper expands the single pairwise sorting algorithm into a more detailed pair-level parallel sorting. First, the non-interactive item set is refined into two categories through the concept of frequent item sets: uncertain feedback and negative feedback. Secondly, the algorithm of fusion of BPR and list sorting is used to relax the assumptions of independence, and two sets are analyzed separately. This also alleviates the sparsity problem of implicit data due to natural imbalance. Finally, a simulation experiment was performed on the public data set. The experimental results show that the refined BPR algorithm has a better improvement than the baseline algorithm.
- Gantz J, Reinsel D. The Digital Universe in 2020: Big Data, Bigger Digital Shadows, and Biggest Growth in the Far East. IDC iView: IDC Analyze the Future, 2012. 1-16Google Scholar
- Liu Q, Ge Y, Li Z. Personalized Travel Package Recommendation[C] // IEEE International Conference on Data Mining. IEEE, 2011Google Scholar
- Hongke Zhao, Le Wu, Qi Liu, Investment Recommendation in P2P Lending: A Portfolio Perspective with Risk Management[C] //2014 IEEE International Conference on Data Mining. IEEE, 2014:1109-1114Google Scholar
- Pan, Jincai, Wang, One-class collaborative filtering based on rating prediction and ranking prediction[J]. Knowledge-based systems, 2017Google Scholar
- Hu Y F, Koren, Y, Volinsky, Chris. Collaborative Filtering for Implicit Feedback Datasets[C] //2008 Eighth IEEE International Conference on Data Mining, December 15-19,2008Google Scholar
- Rendle S, Freudenthaler C, Gantner Z, BPR: Bayesian Personalized Ranking from Implicit Feedback [C] //Pro-ceedings of the 25th Conference on Uncertainty in Artificial Intelligence. New York: ACM, 2009: 452-461Google Scholar
- Chen X, Wang P, Qin Z, HLBPR: A Hybrid Local Bayesian Personal Ranking Method[C] //the 25th International Conference Companion. International World Wide Web Conferences Steering Committee, 2016Google Scholar
- Gantner Z, Drumond L, Freudenthaler C, Bayesian Personalized Ranking for Non-Uniformly Sampled Items[J]. Journal of Machine Learning Research, 2013, 18:231-247Google Scholar
- Runlong Yu, Yunzhou Zhang, Yuyang Ye, Multiple Pairwise Ranking with Implicit Feedback[C]// the 27th ACM International Conference. ACM, 2018Google Scholar
- Song D, Meyer D, Tao D. Efficient Latent Link Recommendation in Signed Networks[C]// Acm Sigkdd International Conference on Knowledge Discovery & Data Mining. ACM, 2015Google Scholar
Recommendations
Typicality-Based Collaborative Filtering Recommendation
Collaborative filtering (CF) is an important and popular technology for recommender systems. However, current CF methods suffer from such problems as data sparsity, recommendation inaccuracy, and big-error in predictions. In this paper, we borrow ideas ...
Using a trust network to improve top-N recommendation
RecSys '09: Proceedings of the third ACM conference on Recommender systemsTop-N item recommendation is one of the important tasks of recommenders. Collaborative filtering is the most popular approach to building recommender systems which can predict ratings for a given user and item. Collaborative filtering can be extended ...
Ranking and Suggesting Popular Items
We consider the problem of ranking the popularity of items and suggesting popular items based on user feedback. User feedback is obtained by iteratively presenting a set of suggested items, and users selecting items based on their own preferences either ...
Comments