Mining Tibetan-Chinese Bilingual Entities from Wikipedia

Entity translation pairs play an important role in NLP applications, such as cross language information retrieval and machine translation. The named entity and domain entity are key factors that affect the performance of the system. However, the entity translations can hardly he found in the present bilingual dictionary or parallel corpus. There are lots of Tibetan new neologisms and named entities in Tibetan Wikipedia, and this paper proposes a new method to automatically mining method of Tibetan and Chinese bilingual entity translation from Wikipedia based on the language interlink and page feature. We construct an extract pattern of Tibetan and Chinese entity translation pairs gained from the previous work, and adopt multi-feature candidate translation pairs to distinguish the selection model. The results verify that the entity translation mining method can achieve high accuracy.