Linguistic Description and Automatic Extraction of Definitions from German Court Decisions

Abstract

This paper discusses the use of computational linguistic technology to extract definitions from a large corpus of German court decisions. We present a corpus-based survey of definition structures used in this kind of document. We then evaluate the results of a definition extraction system that uses patterns identified in this survey to extract from dependency parsed text. We show how an automatically induced ranking function improves the quality of the search results of this system, and we discuss methods for the acquisition of further extraction rules.

Resources

Details