Atigeo at TREC 2014 Clinical Decision Support Task
Abstract
The TREC 2014 Clinical Decision Support Track task involves retrieval and ranking of medical journal articles with respect to their relevance to prescribing tests, diagnosing or treating a patient represented in a short case report. The Atigeo xPatterns platform supports a variety of ensemble methods for developing and tuning information retrieval (IR) system components for a task and/or domain using labeled data. For TREC 2014, we combine results from an ensemble of search engines, each with a configurable suite of natural language processing (NLP) components, to compute a relevance score for each article and topic. We describe our ensemble approach, the strategies and tools we use to create labeled data to support this approach, the components in our IR / NLP pipeline, and our results on the TREC 2014 CDS task.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 2014
- Accession Number
- ADA618775
Entities
People
- Alex Thomas
- Chenchieh Hsu
- Joseph F. Mccarthy
- Yishul Wei