To appear in Proc. Second Symposium on Natural Language Processing, SNLP-95, August 2-4, 1995, Bangkok, Thailand. Semantic Analysis in The Mikrokosmos Machine Translation Project

Stephen Beale, Sergei Nirenburg and Kavi Mahesh

Computing Research Laboratory
Box 30001, Dept. 3CRL
New Mexico State University
Las Cruces, NM 88003-0001 USA
sb,sergei,mahesh@crl.nmsu.edu

Abstract:

The Mikrokosmos (uK) Machine Translation System is a knowledge-based machine translation (KBMT) system under development at New Mexico State University. Unlike previous research in interlingual MT, this project is a large-scale, practical MT system. By the end of the year, a lexicon of approximately 20,000 Spanish words (with over 35,000 word senses) supported by an ontology of 6000 concepts will be in place. High quality semantic analyses of over 400 article-length Spanish texts in the domain of company mergers and acquisitions will have been produced. In the coming year, uK intends to expand into other languages, notably Thai.

This paper introduces the central concepts involved in KBMT, including Text-Meaning Representation (TMR), ontology, and the semantic lexicon. The semantic analyzer ``engine'' will then be described in detail, with examples of how knowledge from the ontology, lexicon and syntactic analysis are combined to create the basic semantic dependency structures found in the TMR outputs. Several issues in semantic analysis programming will be briefly probed, including dependency-directed processing, ``best-first'' search, and knowledgeable treatment of unexpected and ambiguous inputs.

Kavi Mahesh
Sun Nov 12 15:24:36 MST 1995