Taxonomy-based Negative Sampling In Personalized Semantic Search for E-commerce

TL;DR

This paper introduces a taxonomy-based negative sampling method for personalized semantic search in e-commerce, improving retrieval relevance, training efficiency, and user engagement by integrating product taxonomy and user behavior.

Contribution

The paper proposes a novel taxonomy-based hard-negative sampling strategy combined with user personalization for improved e-commerce search models.

Findings

Outperforms BM25, ANCE, and neural baselines on Recall@K.

Increases conversion rate, add-to-cart rate, and average order value in live tests.

Reduces training overhead and speeds up convergence.

Abstract

Large retail outlets offer products that may be domain-specific, and this requires having a model that can understand subtle differences in similar items. Sampling techniques used to train these models are most of the time, computationally expensive or logistically challenging. These models also do not factor in users' previous purchase patterns or behavior, thereby retrieving irrelevant items for them. We present a semantic retrieval model for e-commerce search that embeds queries and products into a shared vector space and leverages a novel taxonomy-based hard-negative sampling(TB-HNS) strategy to mine contextually relevant yet challenging negatives. To further tailor retrievals, we incorporate user-level personalization by modeling each customer's past purchase history and behavior. In offline experiments, our approach outperforms BM25, ANCE and leading neural baselines on Recall@K, while live A/B testing shows substantial uplifts in conversion rate, add-to-cart rate, and average order value. We also demonstrate that our taxonomy-driven negatives reduce training overhead and accelerate convergence, and we share practical lessons from deploying this system at scale.