Publikation

A Type-Driven Method for Compacting MMorph Resources

Hans-Ulrich Krieger, Feiyu Xu

In: Proceedings of RANLP 2003. International Conference on Recent Advances in Natural Language Processing (RANLP) 9/2003.

Abstrakt

This paper describes an offline compaction method that removes both redundancies and spurious ambiguities from MMorph lexical databases. The described technique increases the efficiency of systems using MMorph, since it shrinks the size of the lexicons and comes up with fewer readings for a morphological form. Our approach not only is interesting to MMorph, but also to other lexicons, which build on an attribute-value representation of lexical information. The compaction method is part of the SProUT shallow processing system.

compactmmorph.pdf (pdf, 88 KB )

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence