United StatesPublic sector2017 – ongoingNational Library of Medicine
PubMed Best Match machine-learning search ranking
PubMed reported 60% growth in relevance-sorted searches after deploying Best Match
Problem
Rapid growth in biomedical literature made relevant papers harder to find. PubMed needed a stronger alternative to sorting results by publication date.
Approach
NLM's National Center for Biotechnology Information deployed Best Match in June 2017. It trained the ranking model on past searches and article-level relevance signals. The system reranked an initial set of search results using LambdaMART.
Results
Takeaway
The reported growth measured use of relevance sorting. Separate testing measured whether users clicked results.
Categories
- Size
- Large enterprise
- Industry
- Public sector
- Business function
- IT and R&D
- Innovation type
- Product and service innovation
- Adoption
- Built in-house
- Evidence
- Company's own claim
Sources
Source summaries
NLM researchers described the ranking system and its June 2017 deployment. They reported 60% growth in relevance-sorted searches and over 20% higher click-through in real-world testing.
Full details are in the original sources.
