Discriminative Spoken Language Understanding Using Statistical Machine Translation Alignment Methods

In this paper, we study the discriminative modeling of Spo- ken Language Understanding (SLU) using Conditional Random Fields (CRF). Previous discriminative approaches to SLU have been dependent on n-gram features.We have used Statistical Machine Translation (SMT) alignment methods to align the abstract labels, and consider those align- ments as the labels of the aligned words. Using the proposed alignment method and state transition features, the model performance has im- proved. Furthermore, we have compared the proposed method with two baseline approaches; Hidden Vector States (HVS) and baseline-CRF. The results show that for the F-measure the proposed method outperforms HVS by 1:74% and baseline-CRF by 1:7% on ATIS corpus.