Out of Vocabulary (OOV) is one of the major problems in Machine translation and Cross Language Information Retrieval (CLIR). As time goes on, more Chinese new words appear, these new words and their English translations are not collected in the bilingual dictionary used in the CLIR or machine translation system, so they belong to the OOV name entities. From observation, we know Chinese OOV name entities actually can be categorized into multiple types, so in this paper, we propose and implement an Chinese-English OOV translation mining system, and follow the divide and conquer strategy, we further categorize the Chinese OOV name entities into three types: Foreign word, Chinese name, and Chinese abbreviation, and then deal with them separately. When this system is combined with new word mining system, we can collect new name entities for the bilingual dictionary. From the experiment, we can observe that categorizing the Chinese OOV words helps find translations of OOV words and get a decent result.
展开▼