TADT-CSA: Temporal Advantage Decision Transformer with Contrastive State Abstraction for Generative Recommendation

TL;DR

This paper introduces TADT-CSA, a novel generative recommendation model that combines temporal advantage signals with contrastive state abstraction to improve trajectory quality and state representation efficiency.

Contribution

The paper proposes a new TADT-CSA model integrating temporal advantage signals and contrastive state abstraction, enhancing trajectory stitching and state representation in generative recommendation.

Findings

TADT-CSA outperforms baseline models on public datasets.

The model effectively captures long-term returns and sequential trends.

State representations are more expressive and computationally efficient.

Abstract

With the rapid advancement of Transformer-based Large Language Models (LLMs), generative recommendation has shown great potential in enhancing both the accuracy and semantic understanding of modern recommender systems. Compared to LLMs, the Decision Transformer (DT) is a lightweight generative model applied to sequential recommendation tasks. However, DT faces challenges in trajectory stitching, often producing suboptimal trajectories. Moreover, due to the high dimensionality of user states and the vast state space inherent in recommendation scenarios, DT can incur significant computational costs and struggle to learn effective state representations. To overcome these issues, we propose a novel Temporal Advantage Decision Transformer with Contrastive State Abstraction (TADT-CSA) model. Specifically, we combine the conventional Return-To-Go (RTG) signal with a novel temporal advantage (TA) signal that encourages the model to capture both long-term returns and their sequential trend. Furthermore, we integrate a contrastive state abstraction module into the DT framework to learn more effective and expressive state representations. Within this module, we introduce a TA-conditioned State Vector Quantization (TAC-SVQ) strategy, where the TA score guides the state codebooks to incorporate contextual token information. Additionally, a reward prediction network and a contrastive transition prediction (CTP) network are employed to ensure the state codebook preserves both the reward information of the current state and the transition information between adjacent states. Empirical results on both public datasets and an online recommendation system demonstrate the effectiveness of the TADT-CSA model and its superiority over baseline methods.