Evaluation Techniques Applied to Domain Tuning of MT Lexicons

Fazil Ayan
 
University
of Maryland, College Park


Computational Linguistics Colloquium

September 17, 2003, 11:00am, AVW Room 2120


We describe a set of evaluation techniques applied to domain tuning of bilingual lexicons for machine translation.  Our overall objective is to translate a domain-specific document in a foreign language (in this case, Chinese) to English.  First, we perform an intrinsic evaluation of the effectiveness of our domain-tuning techniques by comparing our domain-tuned lexicon to a manually constructed domain-specific bilingual termlist.  Our results indicate that we achieve 66% recall and 95% precision with respect to a human-derived gold standard. Next, an extrinsic evaluation demonstrates that our domain-tuned lexicon improves the Bleu scores 50% over a statistical system -with a smaller improvement when the system is trained on a uniformly-weighted dictionary.

 

For the colloquium series schedule, see the UMD Computational Linguistics Colloquium Series web page at http://umiacs.umd.edu/~resnik/cl_colloquium/. If you are interested in meeting with the speaker, please contact Doug Oard (oard@umiacs.umd.edu).