Non-autoregressive Generative Models for Reranking Recommendation
Open Access
- 24 August 2024
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 5625-5634
Abstract
Contemporary recommendation systems are designed to meet users' needs by delivering tailored lists of items that align with their specific demands or interests. In a multi-stage recommendation system, reranking plays a crucial role by modeling the intra-list correlations among items. The key challenge of reranking lies in the exploration of optimal sequences within the combinatorial space of permutations. Recent research proposes a generator-evaluator learning paradigm, where the generator generates multiple feasible sequences and the evaluator picks out the best sequence based on the estimated listwise score. The generator is of vital importance, and generative models are well-suited for the generator function. Current generative models employ an autoregressive strategy for sequence generation. However, deploying autoregressive models in real-time industrial systems is challenging. Firstly, the generator can only generate the target items one by one and hence suffers from slow inference. Secondly, the discrepancy between training and inference brings an error accumulation. Lastly, the left-to-right generation overlooks information from succeeding items, leading to suboptimal performance. To address these issues, we propose a Non-AutoRegressive generative model for reranking Recommendation (NAR4Rec) designed to enhance efficiency and effectiveness. To tackle challenges such as sparse training samples and dynamic candidates, we introduce a matching model. Considering the diverse nature of user feedback, we employ a sequence-level unlikelihood training objective to differentiate feasible sequences from unfeasible ones. Additionally, to overcome the lack of dependency modeling in non-autoregressive models regarding target items, we introduce contrastive decoding to capture correlations among these items. Extensive offline experiments validate the superior performance of NAR4Rec over state-of-the-art reranking methods. Online A/B tests reveal that NAR4Rec significantly enhances the user experience. Furthermore, NAR4Rec has been fully deployed in a popular video app Kuaishou with over 300 million daily active users.Keywords
All Related Versions
This publication has 21 references indexed in Scilit:
- Personalized re-ranking for recommendation2019
- Deep Interest Network for Click-Through Rate Prediction2018
- xDeepFM2018
- Learning a Deep Listwise Context Model for Ranking Refinement2018
- Relevance Judgments2017
- Accurately Interpreting Clickthrough Data as Implicit FeedbackACM SIGIR Forum, 2017
- Wide & Deep Learning for Recommender Systems2016
- Deep Neural Networks for YouTube Recommendations2016
- Eye tracking and online search: Lessons learned and challenges aheadJournal of the American Society for Information Science and Technology, 2008
- Learning to rank using gradient descent2005