Feature Selection and Class-Weight Tuning Using Genetic Algorithm for Bio-molecular Event Extraction

In this paper we present an efficient method for event extraction in bio-medical text. Event extraction deals with finding more detailed biological phenomenon, which is more challenging compared to simple binary relation extraction like protein-protein interaction (PPI). The task can be thought of as comprising of the following sub-problems, viz., trigger detection, classification and argument extraction. Event trigger detection and classification deal with identification of event expressions from text and classification of them into nine different categories. We use Support Vector Machine (SVM) as the base learning algorithm, which is trained with a diverse set of features that cover both statistical and linguistic characteristics. We develop a feature selection algorithm using Genetic Algorithm (GA) for determining the most relevant set of features. Experiments on benchmark datasets of BioNLP-2011 shared task datasets show the recall, precision and F-measure values of 48.17%, 65.86% and 55.64% respectively.