M. Siu and M. Ostendorf,
"Integrating a Context-Dependent Phrase Grammar in the variable n-gram framework,"
Proc. ICASSP, to appear.

Learning and modeling multi-word lexical units (phrases) has been the subject of considerable interest in recent years. In this paper, we focus on the learning of phrases and how to model them within the variable n-gram frame. We introduced the notion of context-dependent phrases and suggest one algorithm for unsupervised learning of phrases. Also, we proposed an approach to integrate phrase grammar and the variable n-gram without the need of explicitly handling multi-word lexical items. The combined variable n-gram phrase grammar improves recognition accuracy on the Switchboard corpus over both the baseline trigram and using variable n-gram alone.

Get Postscript

Return to SSLI Lab Publications