Identification of V+N compound nouns in Chinese-English machine translation based on rules in patent texts

We address the problem of exclusion of verbs in sentences which are not predicates in Machine Translation and V+N compound nouns are the most apparent phenomena. We propose several identification rules based on HNC theory to lower the error rate of predicate-identification and verb-translation. Result is got by testing in existing MT system, which has a precision of 83.22% and a recall of 65.56%. And the result is valid to the identification and translation of Eigen Chunk¹.