Automatic extraction of clause relationships from a treebank

Abstract

The paper concentrates on deriving non-obvious information about clause structure of complex sentences from the Prague Dependency Treebank. Individual clauses and their mutual relationship are not explicitly annotated in the treebank, therefore it was necessary to develop an automatic method transforming the original annotation concentrating on the syntactic role of individual word forms into a scheme describing the relationship between individual clauses. The task is complicated by a certain degree of inconsistency in original annotation with regard to clauses and their structure. The paper describes themethod of deriving clause-related information from the existing annotation and its evaluation. © Springer-Verlag Berlin Heidelberg 2009.