Researchers have developed a new system for e-commerce search that enhances item discoverability by generating related user intents. This two-stage architecture uses large language models for common queries and a fine-tuned small language model with LoRA adapters for less frequent queries. The system improves retrieval effectiveness, increasing discovery coverage from 60% to 80% while reducing inference costs by approximately 30%. This approach aims to balance marketplaces by exposing long-tail and emerging products. AI
IMPACT Enhances e-commerce search by improving product discovery and potentially balancing marketplace exposure for long-tail items.
RANK_REASON The cluster contains a research paper detailing a new system for e-commerce search.
Read on arXiv cs.IR (Information Retrieval) →
- arXiv
- e-commerce
- Hugging Face
- information retrieval
- large language models
- LLM-as-a-Judge
- Lora
- small language model
AI-generated summary · Google Gemini · from 3 sources. How we write summaries →
