Corpus-Based Chinese-Korean Abstracting Translation System

Abstract

A Corpus-Based Chinese-Korean Abstracting Translation System is designed and implemented. Firstly, a text indexing method called Natural Hierarchical Network(NHN) is introduced, and then a Corpus-Based Word Segmentation algorithm is developed with the segmentation correctness of 98% for open test. Based on a words weighting function and a sentence importance weighting function which can dynamically calculate the importance of words and sentences by using the word frequency both in corpus and context, word length, sentence length and so on, an abstracting system is implemented to produce abstracts of texts in deferent languages and domains by any abstracting rate. Experiments show that generally abstracts produced by 10% to 20% abstracting rates can cover 90% of the important sentences of the input texts. Finally, combines with an Example-Based Chinese-Korean Machine Translation System, the generated abstracts are translated into target language with the correctness of translation of more than 70% by the important words oriented machine translation strategy.