Knowledge-Based Access to the Bio-Medical Literature, Ontologically-Grounded Experiments for the TREC 2003 Genomics Track

The Tarragon Consulting team participated in the primary task of the TREC 2003 Genomics Track. We used a combination of knowledge-engineering and corpus analysis to construct semantic models of the interactions between genes/proteins and other biological entities in the organism, and then used automatic methods to convert these models into evidential queries that could be executed by the K2 search engine from Verity, Inc. The primary goal of our participation in the Genomics Track was to establish a performance baseline using ontologicallygrounded techniques that are scalable and implementable using current commercial retrieval technology. The results from both our official submissions and subsequent experiments demonstrate that good performance can be achieved using our approach. Overall Approach In our approach we focused on “function” as opposed to the other aspects of the basic biology of the gene and its protein products. That is, we interpreted the primary task to be one tha...

• Type: Conference
• Year: 2003
• Where: TREC
• Authors: Richard Tong, John Quackenbush, Mark Snuffin
• Keywords: Gene/protein Subsystem, Primary Task, Semantic Models, TREC 2003, TREC 2008