Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enhancing Relevance of Embedding-based Retrieval at Walmart

Juexin Lin, Sachin Yadav, Feng Liu, Nicholas Rossi, Praveen R. Suram, Satya Chembolu, Prijith Chandran, Hrushikesh Mohapatra, Tony Lee, Alessandro Magnani, Ciya Liao

TL;DR

This work tackles persistent relevance degradation in Walmart's embedding-based retrieval by introducing a human-feedback relevance reward model (RRM) and integrating it through label revision and a multi-objective training objective. It also enhances robustness to misspellings via typo-aware training, and improves negative sampling with semi-positives and stratified PT-based sampling. Empirical results from offline evaluations and live online A/B tests show that the proposed approaches, especially the multi-objective loss, yield substantial gains in relevance metrics and revenue, culminating in production deployments. The methods collectively address data noise, misspellings, and candidate quality in a large-scale, hybrid retrieval setting, with clear practical impact for e-commerce search quality.

Abstract

Embedding-based neural retrieval (EBR) is an effective search retrieval method in product search for tackling the vocabulary gap between customer search queries and products. The initial launch of our EBR system at Walmart yielded significant gains in relevance and add-to-cart rates [1]. However, despite EBR generally retrieving more relevant products for reranking, we have observed numerous instances of relevance degradation. Enhancing retrieval performance is crucial, as it directly influences product reranking and affects the customer shopping experience. Factors contributing to these degradations include false positives/negatives in the training data and the inability to handle query misspellings. To address these issues, we present several approaches to further strengthen the capabilities of our EBR model in terms of retrieval relevance. We introduce a Relevance Reward Model (RRM) based on human relevance feedback. We utilize RRM to remove noise from the training data and distill it into our EBR model through a multi-objective loss. In addition, we present the techniques to increase the performance of our EBR model, such as typo-aware training, and semi-positive generation. The effectiveness of our EBR is demonstrated through offline relevance evaluation, online AB tests, and successful deployments to live production. [1] Alessandro Magnani, Feng Liu, Suthee Chaidaroon, Sachin Yadav, Praveen Reddy Suram, Ajit Puthenputhussery, Sijie Chen, Min Xie, Anirudh Kashi, Tony Lee, et al. 2022. Semantic retrieval at walmart. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 3495-3503.

Enhancing Relevance of Embedding-based Retrieval at Walmart

TL;DR

This work tackles persistent relevance degradation in Walmart's embedding-based retrieval by introducing a human-feedback relevance reward model (RRM) and integrating it through label revision and a multi-objective training objective. It also enhances robustness to misspellings via typo-aware training, and improves negative sampling with semi-positives and stratified PT-based sampling. Empirical results from offline evaluations and live online A/B tests show that the proposed approaches, especially the multi-objective loss, yield substantial gains in relevance metrics and revenue, culminating in production deployments. The methods collectively address data noise, misspellings, and candidate quality in a large-scale, hybrid retrieval setting, with clear practical impact for e-commerce search quality.

Abstract

Embedding-based neural retrieval (EBR) is an effective search retrieval method in product search for tackling the vocabulary gap between customer search queries and products. The initial launch of our EBR system at Walmart yielded significant gains in relevance and add-to-cart rates [1]. However, despite EBR generally retrieving more relevant products for reranking, we have observed numerous instances of relevance degradation. Enhancing retrieval performance is crucial, as it directly influences product reranking and affects the customer shopping experience. Factors contributing to these degradations include false positives/negatives in the training data and the inability to handle query misspellings. To address these issues, we present several approaches to further strengthen the capabilities of our EBR model in terms of retrieval relevance. We introduce a Relevance Reward Model (RRM) based on human relevance feedback. We utilize RRM to remove noise from the training data and distill it into our EBR model through a multi-objective loss. In addition, we present the techniques to increase the performance of our EBR model, such as typo-aware training, and semi-positive generation. The effectiveness of our EBR is demonstrated through offline relevance evaluation, online AB tests, and successful deployments to live production. [1] Alessandro Magnani, Feng Liu, Suthee Chaidaroon, Sachin Yadav, Praveen Reddy Suram, Ajit Puthenputhussery, Sijie Chen, Min Xie, Anirudh Kashi, Tony Lee, et al. 2022. Semantic retrieval at walmart. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 3495-3503.
Paper Structure (24 sections, 6 equations, 3 figures, 4 tables)

This paper contains 24 sections, 6 equations, 3 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. A new labeling framework
  4. Relevance reward model
  5. Label revision
  6. Relevance label
  7. Semantic Model
  8. Model architecture
  9. Loss function
  10. Typos-aware training
  11. Negative sampling schemes
  12. In-batch hard negatives
  13. Offline hard negatives
  14. Stratified sampling
  15. Experiments and Results
  16. ...and 9 more sections

Figures (3)

  • Figure 1: Relevance reward model is trained with multi-class cross entropy loss leveraging the cross-encoder architecture.
  • Figure 2: Embedding-based retrieval model adopts Siamese dual encoder architecture. Query and document have Blue components are shared between query and products.
  • Figure 3: EM Recall@20 with respect to various hyper-parameter $\omega$ values