SRSL - 2009

is to bring together researchers interested in the semantic representation of spoken corpora, especially spontaneous speech. On one hand, the semantic gap between contents conveyed by natural languages and their formal representations is a burning aspect in tasks such as information extraction and corpus annotation. The current state-of-the-art supports solutions from very different backgrounds and perspectives, but still remain important and complex issues to deal with, such as the accurate segmentation of speech in semantic units. The discussion of those aspects are one of the main reasons for this workshop. On the other hand, spoken language is a pending issue in computational linguistics and artificial intelligence, both traditionally focused on written language, although semantic processing of speech is necessary for the understanding of both natural and human-machine interaction. Finally, the problems found when trying to linguistically structure spontaneous speech are leading to works focused on its semantic representation. In-depth research on the semantic representation of speech can provide us with a suitable basis for further analysis of related linguistic levels, like prosody or pragmatics.

• Corpus annotation: structures (frame-banks, proposition banks, etc.) and concepts (ontologies, named entities, etc.).
• Content identification and segmentation in spontaneous speech.
• Semantic interpretation in dialogues.
• Dialogue and discourse structures.
• Topic Detection and tracking.
• Multimodal Representations including speech.
• Natural language understanding and reasoning in spoken dialogue systems.
• Speech in embedded systems.
• Project descriptions about applications of spoken corpora semantic representations.
• Standardisation work.
• Interoperability/Comparison work of spoken and written corpora.