Academic Commons

Presentations (Communicative Events)

Automatic Summarization of Conversational Multi-Party Speech

Galley, Michel

Document summarization has proven to be a desirable component in many information management systems, complementing core information retrieval and browsing functionalities. The use of document summarization techniques is especially important with speech documents such as meeting transcriptions, since they are particularly difficult to navigate. As opposed to their written counterparts, spoken documents lack structural elements like title, headers, and paragraphs that can ease the task of the information seeker. Document summarization can reduce the overhead of navigating large collections of speech data. For example, summarization techniques can be used to extract or highlight relevant passages in a full transcription. Alternatively, similar techniques might be used to generate short text substitutes, or abstracts, that capture the “aboutness” of the meeting, while discarding disfluencies and other unimportant elements.


More About This Work

Academic Units
Computer Science
AAAI'06 proceedings of the 21st national conference on Artificial intelligence
Published Here
June 30, 2013