Adnan KhanAdnanProduct & AI
San Francisco Bay Area
Get in touch

Source note3 min read

Improving Item Discoverability in e-Commerce Search via Related Intent Generation

Source

arXiv

arxiv.org/abs/2607.27172v1

Read the source

Source

  • Source name: arXiv
  • Source kind: paper
  • Authors / organization: Ji Xin, Xiao Xiao, Ishan Bhatt, Vinesh Gudla, Trace Levinson, Raochuan Fan, Shishir Kumar Prasad, Prakash Putta, Tejaswi Tenneti
  • Venue: arXiv: cs.IR
  • Published: 2026-07-29
  • Collected: 2026-07-30
  • URL: http://arxiv.org/abs/2607.27172v1
  • PDF: https://arxiv.org/pdf/2607.27172v1
  • DOI: UNKNOWN

Plain-Language Brief

The researchers present a scalable system called 'discovery-augmented search' that aims to improve the discoverability of items in e-commerce marketplaces, particularly grocery. This system uses intent-conditioned recall expansion and a two-stage hybrid architecture involving large and small language models. The results show an improvement in both intent generation quality and downstream retrieval effectiveness, extending discovery coverage from approximately 60% to 80% of query traffic at roughly 30% of the teacher model's inference cost.

What the Document Says

  • The paper proposes a system for discovery-augmented search.
  • The system is designed for e-commerce marketplaces, particularly grocery.
  • Traditional search systems prioritize precision over recall and are limiting in these markets.
  • The proposed system leverages intent-conditioned recall expansion to improve discoverability.
  • The system generates implicit user intents to expand candidate recall while maintaining relevance.
  • A two-stage hybrid architecture is used: a large language model for head queries and a finetuned small language model for tail queries.
  • The small language model is trained via LoRA adapters and teacher-student distillation.
  • The system improves both intent generation quality and downstream retrieval effectiveness.
  • Discovery coverage is extended from approximately 60% to 80% of query traffic at roughly 30% of the teacher model's inference cost.
  • The results demonstrate the viability of this approach for deployment in large-scale marketplaces.

Key Concepts Mentioned

  • e-commerce search
  • item discoverability
  • intent-conditioned recall expansion
  • large language models (LLMs)
  • small language model (SLM)
  • LoRA adapters
  • teacher-student distillation

Claims Made by the Document

  • Traditional search systems are limiting in e-commerce marketplaces.
  • The proposed system improves both intent generation quality and downstream retrieval effectiveness.
  • Discovery coverage is extended from approximately 60% to 80% of query traffic at roughly 30% of the teacher model's inference cost.

Evidence Used

  • User satisfaction and commercial outcomes depend heavily on the discoverability of substitute, complementary, and thematically related items.
  • Results demonstrate that our approach improves both intent generation quality and downstream retrieval effectiveness, extending discovery coverage from approximately 60% to 80% of query traffic at roughly 30% of the teacher model's inference cost.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Traditional search systems are optimized to retrieve items that strictly match a query, often prioritizing precision over recall. In e-commerce marketplaces and particularly grocery, this paradigm is limiting, as user satisfaction and commercial outcomes depend heavily on the discoverability of substitute, complementary, and thematically related items. In this paper, we present a scalable system for discovery-augmented search that leverages intent-conditioned recall expansion. Our approach generates implicit user intents to expand candidate recall while maintaining relevance. The system addresses the cost-quality tradeoff of generative retrieval through a two-stage hybrid architecture. First, we leverage closed-weight large language models (LLMs) to maximize discoverability for head queries. To extend these benefits to tail queries, we then introduce a finetuned small language model (SLM), trained via LoRA adapters and teacher-student distillation. We evaluate the system using a rigorous dual framework: (a) LLM-as-a-judge metrics validated against human preferences for semantic quality, and (b) end-to-end session-level purchase analysis. Results demonstrate that our approach improves both intent generation quality and downstream retrieval effectiveness, extending discovery coverage from approximately 60% to 80% of query traffic at roughly 30% of the teacher model's inference cost, offering a viable path for deployment in large-scale marketplaces. Beyond relevance gains, discovery-augmented search may serve as a marketplace-balancing mechanism, giving long-tail and emerging supply an opportunity for query-conditioned exposure.