Distinguished Quantized Guidance for Diffusion-based Sequence Recommendation

TL;DR

The paper tackles diffusion-model-based sequential recommendation by addressing two core problems: heterogeneous/noisy guidance from sparse histories and biased generation that overemphasizes popular items. It proposes DiQDiff, which combines Semantic Vector Quantization (SVQ) to extract robust, semantically rich guidance and Contrastive Discrepancy Maximization (CDM) to enforce distinct, personalized generations within a conditional DDPM framework. SVQ uses a discrete codebook to quantize sequences into semantic vectors, while CDM introduces a contrastive objective to push denoised outputs apart across different user-guidance signals, mitigating popularity bias. Empirical results on four public datasets show state-of-the-art performance, with notable improvements on large-scale data, demonstrating the method's effectiveness and potential for broader application in personalized generativeRecommendation systems.

Abstract

Diffusion models (DMs) have emerged as promising approaches for sequential recommendation due to their strong ability to model data distributions and generate high-quality items. Existing work typically adds noise to the next item and progressively denoises it guided by the user's interaction sequence, generating items that closely align with user interests. However, we identify two key issues in this paradigm. First, the sequences are often heterogeneous in length and content, exhibiting noise due to stochastic user behaviors. Using such sequences as guidance may hinder DMs from accurately understanding user interests. Second, DMs are prone to data bias and tend to generate only the popular items that dominate the training dataset, thus failing to meet the personalized needs of different users. To address these issues, we propose Distinguished Quantized Guidance for Diffusion-based Sequence Recommendation (DiQDiff), which aims to extract robust guidance to understand user interests and generate distinguished items for personalized user interests within DMs. To extract robust guidance, DiQDiff introduces Semantic Vector Quantization (SVQ) to quantize sequences into semantic vectors (e.g., collaborative signals and category interests) using a codebook, which can enrich the guidance to better understand user interests. To generate distinguished items, DiQDiff personalizes the generation through Contrastive Discrepancy Maximization (CDM), which maximizes the distance between denoising trajectories using contrastive loss to prevent biased generation for different users. Extensive experiments are conducted to compare DiQDiff with multiple baseline models across four widely-used datasets. The superior recommendation performance of DiQDiff against leading approaches demonstrates its effectiveness in sequential recommendation tasks.

Figures (7)

• Figure 1: Challenges in adapting DMs to the sequential recommendation: (left) the heterogeneous (e.g., sparse) or noisy (e.g., mis-click) sequences as guidance, and (right) the biased generation in the item embedding space.
• Figure 2: The framework of DiQDiff. The Semantic Vector Quantization is applied to quantize sequences with a semantic codebook, extracting accurate and robust guidance. The Contrastive Discrepancy Maximization is utilized to maximize the distance between different denoising trajectories, enabling distinguished item generation for different users.
• Figure 3: Quantization and updating process in SVQ.
• Figure 4: The T-SNE visualization of the generated item embeddings on the Toys dataset.
• Figure 5: The T-SNE visualization displays the discrete code vectors in a codebook with $M=32$ on the Toys dataset.