An Approach to Estimate Perplexity Values for Language Models Based on Phrase Classes

In this work we propose an approach to estimate perplexity values for complex language models such as a language model based on phrase classes. The perplexity values obtained by using this method are compared to other typically employed approaches and to the perplexity obtained without any simplification. Experiments over two different corpora were carried out and it can be concluded that the proposed approach provides a good estimation of the perplexity while reduces the computational cost.