摘要
本文研究了N元文法(N gram)统计语言模型中的4种插值平滑算法,在中文语言模型中进行了应用,从语言模型复杂度的角度比较了该4种方法解决零概率问题的效率。
This paper describes interpolation smoothing techniques for statistical language models and their application to language model in China.From the point of view of perplexity of language model,this paper discusses the efficiency of four kinds of methods on probabilities of sparse date.
出处
《信息工程大学学报》
2002年第4期13-15,共3页
Journal of Information Engineering University
