Reproduce “Intent-Aware Neural Query Reformulation for Behavior-Aligned Product Search” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Jayanth Yetukuri, Ishita Khan
eBay Inc. researchers developed an intent-aware neural query reformulation framework for e-commerce search, which leverages behavioral signals to generate query rewrites tailored to distinct buyer intent categories. The Transformer model, especially when considering multiple outputs, demonstrated superior performance in rewrite type coverage, achieving a coverage of 1.00, recall of 0.45, and precision of 0.54, thereby improving search relevance and user engagement.
Key insights
- Leveraging rich, multi-session buyer behavioral signals (in-session and cross-session co-engagement) is crucial for constructing a high-fidelity dataset that captures nuanced user intent for query reformulation.
- Explicitly conditioning neural query reformulation models on distinct buyer intent categories allows for adaptive strategies that optimize specific outcomes, ranging from precision-focused recall enhancement to broad exploratory discovery.
- A unified framework that combines intent-preserving and intent-diversifying reformulation capabilities can bridge traditional retrieval with modern recommendation paradigms, enhancing both search coverage and user engagement.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.