Reducing sparsity improves the recognition of implicit discourse relations

Abstract

The earliest work on automatic detection of implicit discourse relations relied on lexical features. More recently, researchers have demonstrated that syntactic features are superior to lexical features for the task. In this paper we re-examine the two classes of state of the art representations: syntactic production rules and word pair features. In particular, we focus on the need to reduce sparsity in instance representation, demonstrating that different representation choices even for the same class of features may exacerbate sparsity issues and reduce performance. We present results that clearly reveal that lexicalization of the syntactic features is necessary for good performance. We introduce a novel, less sparse, syntactic representation which leads to improvement in discourse relation recognition. Finally, we demonstrate that classifiers trained on different representations, especially lexical ones, behave rather differently and thus could likely be combined in future systems.