TR-S-0014 :2001.2.21

Ruiqiang Zhang

Language Modeling and Statistical Machine Translation

Abstract:Two parts are contained in this report. The first is to report the latest results of language modeling in speech recognition. Detailed information including local N-gram, long distance constraints and linguistic question triggers are integrated in language models by maximum entropy approach. The experimental results prove our models are effective. The second is to discuss our statistical Ngram translation model. Perplexity test was made to evaluate effectiveness of the proposed Ngram translation models.