Disfluency Detection Using Multi-step Stacked Learning

In this paper, we propose a multi-step stacked learning model for disfluency detection. Our method incorporates refined n-gram features step by step from different word sequences. First, we detect filler words. Second, edited words are detected using n-gram features extracted from both the original text and filler filtered text. In the third step, additional n-gram features are extracted from edit removed texts together with our newly induced in-between features to improve edited word detection. We useMax-MarginMarkov Networks (MNs) as the classifier with the weighted hamming loss to balance precision and recall. Experiments on the Switchboard corpus show that the refined n-gram features from multiple steps and MNs with weighted hamming loss can significantly improve the performance. Our method for disfluency detection achieves the best reported F-score 0.841 without the use of additional resources.1