Recently, the focus in the BioNLP domain has shifted from binary relations to more expressive event representations, largely owing to the international popularity of the BioNLP Shared Task (ST) of 2009. This year, the ST'11 provides a further generalization on three key aspects: text type, subject domain, and targeted event types. One of the supporting tasks established to provide more fine-grained text predictions is the extraction of entity relations. We have implemented an extraction system for such non-causal relations between named entities and domain terms, applying semantic spaces and machine learning techniques. Our system ranks second of four participating teams, achieving 37.04% precision, 47.48% recall and 41.62% F-score.
[1]
S. Dongen.
Graph clustering by flow simulation
,
2000
.
[2]
Jun'ichi Tsujii,et al.
Corpus annotation for mining biomedical events from literature
,
2008,
BMC Bioinformatics.
[3]
Sampo Pyysalo,et al.
Integration of Static Relations to Enhance Event Extraction from Text
,
2010,
BioNLP@ACL.
[4]
Sampo Pyysalo,et al.
Overview of the Entity Relations (REL) supporting task of BioNLP Shared Task 2011
,
2011,
BioNLP@ACL.
[5]
Keith Stevens,et al.
The S-Space Package: An Open Source Package for Word Space Models
,
2010,
ACL.