Date of Original Version
Proceedings of International Joint Conference on Natural Language Processing (IJCNLP)
Abstract or Description
This paper presents a graph-based model that integrates prosodic features into an unsupervised speech summarization framework without any lexical information. In particular it builds on previous work using mutually reinforced random walks, in which a two-layer graph structure is used to select the most salient utterances of a conversation. The model consists of one layer of utterance nodes and another layer of prosody nodes. The random walk algorithm propagates scores between layers to use shared information for selecting utterance nodes with highest scores as summaries. A comparative evaluation of our prosody-based model against several baselines on a corpus of academic multi-party meetings reveals that it performs competitively on very short summaries, and better on longer summaries according to ROUGE scores as well as the average relevance of selected utterances.
Proceedings of International Joint Conference on Natural Language Processing (IJCNLP), 648-654.