Event Extraction for Italian Using a Cascade of Finite-State Grammars

Abstract

This paper reports on our experience of adapting a real-world live event extraction system based on a cascade of finite-state extraction grammars to the processing of a new language, namely Italian. The real-time event extraction processing chain and the pattern specification language are briefly presented. The major part of the paper focuses on the creation of event extraction grammars and related resources for English and their adaptation for extracting events in Italian news articles. Some interesting phenomena which complicate the event extraction task for Italian are pinpointed and the results of the evaluation are presented. In particular, we compared two versions of the system for Italian, one based on surface-level patterns and a hybrid one, which integrates slightly more linguistically sophisticated patterns for covering a rich variety of morphological and syntactic constructions in Italian.