Zone identification in biology articles as a basis for information extraction

Abstract

Information extraction (IE) in the biomedical domain is now regarded as an essential technique for the dynamic management of factual information contained in archived journal articles and abstract collections. We aim to provide a technique serving as a basis for pinpointing and organizing factual information related to experimental results. In this paper, we enhance the idea proposed in (Mizuta and Collier, 2004); annotating articles in terms of rhetorical zones with shallow nesting. We give a qualitative analysis of the zone identification (ZI) process in biology articles. Specifically, we illustrate the linguistic and other features of each zone based on our investigation of articles selected from four major online journals. We also discuss controversial cases and nested zones, and ZI using multiple features. In doing so, we provide a stronger theoretical and practical support for our framework toward automatic ZI.