clip

description
   Goal
   Approach
   Datasets
   References

demos

people

contact

Approach

We take an interlingual approach to machine translation. In the analysis phase, Chinese sentences are segmented, parsed, and submitted to semantic interpretation to produce language-independent lexical conceptual structure (LCS) representations. These LCS represntations are then used as the input to the generation phase, which performs lexical selection and syntactic realization, using selected components from ISI's Nitrogen generation system, particularly its statistical preference mechanism for optimizing the fluency of English output.
At present, natural language ambiguity is resolved heuristically, except for the statistical component from Nitrogen; future plans include research on statistical disambiguation at all stages.

previous    home    next

 

Copyright © 1999
Computational Linguistics and Information Processing Laboratory
University of Maryland at College Park, MD 20742, USA

Last modified December 2, 1999

University of Maryland