A pilot study of a predicate-based vector space model for a biomedical search engine

  • Authors:

  • Myungjae Kwak;Gondy Leroy;Jesse D. Martinez

  • Affiliations:

  • School of Information Technology, Macon State College, Macon, GA, U.S.A;School of Information Systems and Technology, Claremont Graduate University, Claremont, CA, U.S.A;Cell Biology and Anatomy, Radiation, Oncology University of Arizona Cancer Center, Tucson, AZ, U.S.A

  • Venue:
  • BIBMW '11 Proceedings of the 2011 IEEE International Conference on Bioinformatics and Biomedicine Workshops
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

A search engine that supports finding precise biomedical statements, but also complementary, and contrasting information would greatly help biomedical researchers. We propose the use of predicates in a search engine's underlying data structure to accomplish this. A predicate is a triple that combines two phrases with a predicate. We report on the development and evaluation of a search engine that includes the predicates in its underlying data structure. The evaluation of the search engine was conducted by comparing three different approaches: keyword-based search, triple-based search, and an additive search that combines keywords and predicates. Cancer researchers provided the queries, relevant to their ongoing work, and evaluated the outcome in a double-blind fashion. The results showed that the combined approach, which combines triple-based and keyword-based approaches, always outperformed the 2 other approaches.