Syntactic Dependency for relation extraction from Biomedical literature

xiaomei wei,jianyong wang,yang li
DOI: https://doi.org/10.2991/citcs.2012.186
2012-01-01
Abstract:Relation extraction is important to improve complex natural language processing (NLP) applications. The Bio-Event extraction in GE shared task is important to understand biological processes. Although some progress has made in GE research, there is still much work to do to improve the performance of the extraction system. In this paper, we build a extraction system based on syntactic dependency technique. Relying on the output of the sentence parsing, we get rich features to build a classification model to separate the candidate edges into positive class and negative class. We obtain promising results on GE develop data set. Especially the results of simple events are comparable with the state-of-the-art GE extraction systems. Keywords-syntactic parsing; Dependency; Classification; extraction.
What problem does this paper attempt to address?