SGER: Discovering and Using the Knowledge in Biology Text

Northeastern University, Boston MA

Investigators

Abstract

A natural language-based system will be built to answer queries based on a 200 million-word body of literature consisting of all the articles from the ASM journals for the last five years. The study consists of proof-of-concept research, so it is experimental in nature. There are two ideas that are key to the analysis, which has the potential to revolutionize the use of electronic literature in biology. The first idea is that biology should be viewed as a collection of natural language structures, containers, which have a standardized form and occur with high frequency in the literature. The second idea is that there is a simple but systematic relation between questions that users ask and the data in the containers found in the text. This will allow the system to answer queries without deep analysis. New technologies, especially visualization tools, will be developed in a uniform environment using Java.