Conceptual biology research supporting platform: Current design and future directions

Abstract

Conceptual biology utilizes a vast amount of published biomedical data to enhance and speed up biomedical research. Current computational study on conceptual biology focuses on hypothesis generation from biomedical literature. Most of the algorithms for hypothesis generation are dedicated to produce one type of hypothesis called pairwise relation by interacting with certain search engines such as PubMed. In order to fully realize the potential of conceptual biology, we designed and implemented a conceptual biology research support platform that consists of a set of interrelated information extraction, mining, reasoning, and visualizing technologies to automatically generate several types of biomedical hypotheses and to facilitate researchers in validating generated hypotheses. In this chapter, we provide detailed descriptions of the platform architecture, the algorithms for generating novel hypotheses, and the technologies for visualizing generated hypotheses. Furthermore, we propose a set of computational procedures and measures for evaluating generated hypotheses. The experimental analysis of the proposed hypothesis generation algorithms is also presented. © 2008 Springer-Verlag Berlin Heidelberg.