Identifying the epistemic value of discourse segments in biology texts

• Authors: Anita de Waard;Paul Buitelaar;Thomas Eigner
• Venue: IWCS-8 '09 Proceedings of the Eighth International Conference on Computational Semantics
• Year: 2009

To manage the flood of information that threatens to engulf (life-)scientists, an abundance of computer-aided tools are being developed. These tools aim to provide access to the knowledge conveyed within a collection of research papers, without actually having to read the papers. Many of these tools focus on text mining, by looking for specific named-entities that have scientific meaning, and relationships between these. An overview of the current state of the art is given in Rebholz-Schuhmann et al. (2005) and Couto et al. (2003). Typically, these tools identify a list of sentences containing relationships between two specific named-entities that can be found using rules or a thesaurus of synonyms. These sentences represent an overview of the interactions that are known with a specific entity, thus precluding the need for an exhaustive literature study. For example, the following are a few sentences that have been found using a typical text mining tool for the relationship 'p53 activates*': 1. The p53 tumor suppressor protein exerts most of its anti-tumorigenic activity by transcriptionally activating several pro-apoptotic genes. 2. We found that p53 ... activates[,] the promoter of the myosin VI gene.