Rethinking Purity and Diversity in Multi-Behavior Sequential Recommendation from the Frequency Perspective
This work addresses noise issues in recommendation systems for users with multiple behaviors, offering an incremental improvement by balancing purity and diversity from a frequency perspective.
The paper tackles the problem of noise in multi-behavior sequential recommendation by re-evaluating the role of high-frequency information, showing that low frequency corresponds to user interest purity and high frequency to diversity, and proposes a model that balances these aspects to improve recommendation performance, achieving effectiveness in experiments on real-world datasets.
In recommendation systems, users often exhibit multiple behaviors, such as browsing, clicking, and purchasing. Multi-behavior sequential recommendation (MBSR) aims to consider these different behaviors in an integrated manner to improve the recommendation performance of the target behavior. However, some behavior data will also bring inevitable noise to the modeling of user interests. Some research efforts focus on data denoising from the frequency domain perspective to improve the accuracy of user preference prediction. These studies indicate that low-frequency information tends to be valuable and reliable, while high-frequency information is often associated with noise. In this paper, we argue that high-frequency information is by no means insignificant. Further experimental results highlight that low frequency corresponds to the purity of user interests, while high frequency corresponds to the diversity of user interests. Building upon this finding, we proposed our model PDB4Rec, which efficiently extracts information across various frequency bands and their relationships, and introduces Boostrapping Balancer mechanism to balance their contributions for improved recommendation performance. Sufficient experiments on real-world datasets demonstrate the effectiveness and efficiency of our model.