Hierarchical Query Classification in E-commerce Search (WWW 2024)
This entry indexes the publication “Hierarchical Query Classification in E-commerce Search,” presented at WWW 2024 (The Web Conference) and published by Amazon Science.
- Official page: https://www.amazon.science/publications/hierarchical-query-classification-in-e-commerce-search
- Venue: WWW 2024
- Topic area: Query Understanding / Verticals, e-commerce search
- Recall routing — selecting which category-specific retrieval indices or verticals to query
- Ranking features — providing category-match signals to the ranking model
- Browse and navigation — powering category filters and guided discovery
- Aligned Query Expansion: Efficient Query Expansion for Information Retrieval
- Beyond the Limitation of a Single Query: Train Your LLM for Query Expansion
- Decomposing Complex Queries for Tip-of-the-tongue Retrieval
- Few-Shot Generative Conversational Query Rewriting (SIGIR 2020, DOI: 10.1145/3397271.3401323)
- Hypothetical Documents or Knowledge Leakage? Rethinking LLM-based Query Expansion
- LLM-Based Query Expansion with Gaussian Kernel Semantic Enhancement
Why hierarchical query classification matters
In e-commerce search, a user query such as a short product phrase must be mapped to a product category. Real production taxonomies are deep and hierarchical: predicting a leaf category alone is brittle, while predicting only coarse categories loses the granularity needed for downstream systems. Hierarchical query classification models the task jointly across taxonomy levels, so that predictions remain consistent with the category tree.
Typical use cases for such classifiers in an industrial search stack include:
Context in query understanding research
This work sits alongside a broader line of research on improving how search systems interpret short, ambiguous queries:
Notes for readers
This page is a curated index entry based on the publication's public metadata. For the full method description, model architecture, training data details, and quantitative results, refer to the original Amazon Science publication page and the paper PDF linked there. Quantitative claims should be verified against the primary source before citation.