Event role extraction using domain-relevant word representations

Abstract

The efficiency of Information Extraction systems is known to be heavily influenced by domain-specific knowledge but the cost of developing such systems is considerably high. In this article, we consider the problem of event extraction and show that learning word representations from unlabeled domain-specific data and using them for representing event roles enable to outperform previous state-of-the-art event extraction models on the MUC-4 data set.