TY - GEN
T1 - Apertium-IceNLP
T2 - 15th International Conference of the European Association for Machine Translation, EAMT 2011
AU - Brandt, Martha Dís
AU - Loftsson, Hrafn
AU - Sigurþórsson, Hlynur
AU - Tyers, Francis M.
PY - 2011
Y1 - 2011
N2 - We describe the development of a prototype of an open source rule-based Icelandic→English MT system, based on the Apertium MT framework and IceNLP, a natural language processing toolkit for Icelandic. Our system, Apertium-IceNLP, is the first system in which the whole morphological and tagging component of Apertium is replaced by modules from an external system. Evaluation shows that the word error rate and the position in dependent word error rate for our prototype is 50.6% and 40.8%, respectively. As expected, this is higher than the corresponding error rates in two publicly available MT systems that we used for comparison. Contrary to our expectations, the error rates of our prototype is also higher than the error rates of a comparable system based solely on Apertium modules. Based on error analysis, we conclude that better translation quality may be achieved by replacing only the tagging component of Apertium with the corresponding module in IceNLP, but leaving morphological analysis to Apertium.
AB - We describe the development of a prototype of an open source rule-based Icelandic→English MT system, based on the Apertium MT framework and IceNLP, a natural language processing toolkit for Icelandic. Our system, Apertium-IceNLP, is the first system in which the whole morphological and tagging component of Apertium is replaced by modules from an external system. Evaluation shows that the word error rate and the position in dependent word error rate for our prototype is 50.6% and 40.8%, respectively. As expected, this is higher than the corresponding error rates in two publicly available MT systems that we used for comparison. Contrary to our expectations, the error rates of our prototype is also higher than the error rates of a comparable system based solely on Apertium modules. Based on error analysis, we conclude that better translation quality may be achieved by replacing only the tagging component of Apertium with the corresponding module in IceNLP, but leaving morphological analysis to Apertium.
UR - http://www.scopus.com/inward/record.url?scp=84857535361&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:84857535361
SN - 9789081486118
T3 - Proceedings of the 15th International Conference of the European Association for Machine Translation, EAMT 2011
SP - 217
EP - 224
BT - Proceedings of the 15th International Conference of the European Association for Machine Translation, EAMT 2011
Y2 - 30 May 2011 through 31 May 2011
ER -