A maximum entropy approach for spoken Chinese understanding

Abstract

In this paper, we present a spoken language understanding method based on the maximum entropy model. We first extract certain features from the corpus, and then train the maximum entropy model with an annotated corpus. We use this model to analyze spoken Chinese into semantic frames. Experiments show that the model can work effectively. © Springer-Verlag Berlin Heidelberg 2003.