e-space
Manchester Metropolitan University's Research Repository

Something old, something new: Identifying knowledge source in bio-events

Nawaz, Raheel and Thompson, Paul and Ananiadou, Sophia (2013) Something old, something new: Identifying knowledge source in bio-events. International Journal of Computational Linguistics and Applications, 4 (1). pp. 129-144. ISSN 0976-0962

[img]
Preview

Download (79kB) | Preview

Abstract

Locating new experimental knowledge in biomedical texts is important for several tasks undertaken by biologists. Although several systems can distinguish between new and existing knowledge, this generally happens at the text zone level. In contrast to text zones, bio-events constitute structured representations of biomedical knowledge. They bridge text with domain knowledge and can be used to develop sophisticated semantic search systems. Typically, event extraction systems locate and classify events and their arguments, but ignore interpretative information (meta-knowledge) from their textual context. Since several events (often nested) can occur in a sentence, determining which event(s) are affected by which textual clues can be complex. We have analysed knowledge source annotation in two bio-event corpora: GENIA-MK (abstracts) and FP-MK (full papers), and have developed a system to classify bioevents automatically according to their knowledge source. Our system performs with an accuracy of over 99% on both abstracts and full papers.

Impact and Reach

Statistics

Downloads
Activity Overview
1Download
50Hits

Additional statistics for this dataset are available via IRStats2.

Actions (login required)

Edit Item Edit Item