An English part-of-speech tagger for machine translation in business domain

Abstract

Part-of-speech tagging is a crucial preprocessing step for machine translation. Current studies mainly focus on the methods, linguistic, statistic, machine learning or hybrid. But so far not many serious attempts have been performed to test the reported accuracy of taggers on different, perhaps domain-specific, corpora. Therefore, this paper presents an English POS tagger for English-Chinese machine translation in business domain, demonstrating how a present tagger can be adapted to learn from a small amount of data and handle unknown words for the purpose of machine translation. A small size of 998k English annotated corpus in business domain is built semi-automatically based on a new tagset, the maximum entropy model is adopted and rule-based approach is used in post-processing. Experiments show that our tagger achieves an accuracy of 99.08% in closed test and 98.14% in open test, which is a quite satisfactory result, compared with the reported best open test result of 97.18% of Stanford English tagger.