Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generative Retrieval with Preference Optimization for E-commerce Search

Mingming Li, Huimu Wang, Zuxu Chen, Guangtao Nie, Yiming Qiu, Guoyu Tang, Lin Liu, Jingwei Zhuo

TL;DR

This framework is designed to effectively learn and align an autoregressive model with target data, subsequently generating the final item through constraint-based beam search and employs a constrained search method to identify key spans for retrieving the final item, thereby enhancing result interpretability.

Abstract

Generative retrieval introduces a groundbreaking paradigm to document retrieval by directly generating the identifier of a pertinent document in response to a specific query. This paradigm has demonstrated considerable benefits and potential, particularly in representation and generalization capabilities, within the context of large language models. However, it faces significant challenges in E-commerce search scenarios, including the complexity of generating detailed item titles from brief queries, the presence of noise in item titles with weak language order, issues with long-tail queries, and the interpretability of results. To address these challenges, we have developed an innovative framework for E-commerce search, called generative retrieval with preference optimization. This framework is designed to effectively learn and align an autoregressive model with target data, subsequently generating the final item through constraint-based beam search. By employing multi-span identifiers to represent raw item titles and transforming the task of generating titles from queries into the task of generating multi-span identifiers from queries, we aim to simplify the generation process. The framework further aligns with human preferences using click data and employs a constrained search method to identify key spans for retrieving the final item, thereby enhancing result interpretability. Our extensive experiments show that this framework achieves competitive performance on a real-world dataset, and online A/B tests demonstrate the superiority and effectiveness in improving conversion gains.

Generative Retrieval with Preference Optimization for E-commerce Search

TL;DR

This framework is designed to effectively learn and align an autoregressive model with target data, subsequently generating the final item through constraint-based beam search and employs a constrained search method to identify key spans for retrieving the final item, thereby enhancing result interpretability.

Abstract

Generative retrieval introduces a groundbreaking paradigm to document retrieval by directly generating the identifier of a pertinent document in response to a specific query. This paradigm has demonstrated considerable benefits and potential, particularly in representation and generalization capabilities, within the context of large language models. However, it faces significant challenges in E-commerce search scenarios, including the complexity of generating detailed item titles from brief queries, the presence of noise in item titles with weak language order, issues with long-tail queries, and the interpretability of results. To address these challenges, we have developed an innovative framework for E-commerce search, called generative retrieval with preference optimization. This framework is designed to effectively learn and align an autoregressive model with target data, subsequently generating the final item through constraint-based beam search. By employing multi-span identifiers to represent raw item titles and transforming the task of generating titles from queries into the task of generating multi-span identifiers from queries, we aim to simplify the generation process. The framework further aligns with human preferences using click data and employs a constrained search method to identify key spans for retrieving the final item, thereby enhancing result interpretability. Our extensive experiments show that this framework achieves competitive performance on a real-world dataset, and online A/B tests demonstrate the superiority and effectiveness in improving conversion gains.
Paper Structure (15 sections, 5 equations, 3 figures, 3 tables)

This paper contains 15 sections, 5 equations, 3 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Approach
  3. Task Re-definition
  4. Supervised Fine-tuning
  5. Preferences Optimization
  6. Constrained Beam-search
  7. Experiments
  8. Datasets and Metrics
  9. Baselines
  10. Implementation Details
  11. Experiment Results
  12. Impact of Different Tasks
  13. Impact of Beam Size
  14. Online A/B Test
  15. Conclusion

Figures (3)

  • Figure 1: The framework of GenR-PO. It comprises four key stages: 1) Task re-definition stage; 2) Supervised fine-tuning stage; 3) Preference alignment stage; and 4) Inference stage based on constrained beam-search.
  • Figure 2: The distribution of percentages across different queries.
  • Figure 3: The performance of different beam sizes.