Researchers have introduced QGDPO, a new method for document expansion in e-commerce search that utilizes Direct Preference Optimization (DPO). This approach aims to improve the quality of generated queries by reducing hallucinations and repetitive content, which are common issues with existing Doc2Query techniques. QGDPO fine-tunes a sequence-to-sequence model and then uses a relevance model to score and filter predictions, effectively removing irrelevant and redundant content. The system has been deployed in production, showing substantial improvements in relevance and user engagement. AI
IMPACT Enhances e-commerce search relevance and user engagement by reducing irrelevant and repetitive query generation.
RANK_REASON The cluster describes a new method presented in an arXiv paper for improving information retrieval in e-commerce search. [lever_c_demoted from research: ic=1 ai=1.0]
Read on arXiv cs.IR (Information Retrieval) →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →