Hi-Gen: Generative Retrieval For Large-Scale Personalized E-commerce Search

TL;DR

Hi-Gen introduces a hierarchical generative retrieval method tailored for large-scale personalized e-commerce search, effectively encoding efficiency and positional information to improve retrieval accuracy and support real-time online deployment.

Contribution

The paper presents Hi-Gen, a novel hierarchical encoding-decoding approach that incorporates efficiency and positional information, advancing generative retrieval in large-scale e-commerce.

Findings

Achieves 3.30% and 4.62% improvements in Recall@1 over SOTA on public and industry datasets.

Introduces a category-guided hierarchical clustering scheme for docID generation.

Develops variants supporting real-time large-scale online recall.

Abstract

Leveraging generative retrieval (GR) techniques to enhance search systems is an emerging methodology that has shown promising results in recent years. In GR, a text-to-text model maps string queries directly to relevant document identifiers (docIDs), dramatically simplifying the retrieval process. However, when applying most GR models in large-scale E-commerce for personalized item search, we must face two key problems in encoding and decoding. (1) Existing docID generation methods ignore the encoding of efficiency information, which is critical in E-commerce. (2) The positional information is important in decoding docIDs, while prior studies have not adequately discriminated the significance of positional information or well exploited the inherent interrelation among these positions. To overcome these problems, we introduce an efficient Hierarchical encoding-decoding Generative retrieval method (Hi-Gen) for large-scale personalized E-commerce search systems. Specifically, we first design a representation learning model using metric learning to learn discriminative feature representations of items to capture semantic relevance and efficiency information. Then, we propose a category-guided hierarchical clustering scheme that makes full use of the semantic and efficiency information of items to facilitate docID generation. Finally, we design a position-aware loss to discriminate the importance of positions and mine the inherent interrelation between different tokens at the same position. This loss boosts the performance of the language model used in the decoding stage. Besides, we propose two variants of Hi-Gen (Hi-Gen-I2I and Hi-Gen-Cluster) to support online real-time large-scale recall in the online serving process. Hi-Gen gets 3.30% and 4.62% improvements over SOTA for Recall@1 on the public and industry datasets, respectively.