Comprehensive Framework for MultiModal Meaning Representation
Abstract
Most of the existing content representation language designs are anchored to a specific modality (such as speech). This paper describes the rationale behind the definition of MMIL, the interface language intended to be used as the exchange format between modules in the MIAMM project, where we try to build a juke box incorporating multiple modalities such as speech, Haptics and Graphics. This interface language has been conceived as a general format for representing multimodal content both at lower (e.g. linguistic analysis) and higher (e.g. communication within the dialogue manager) levels of representations.
Domains
Computation and Language [cs.CL]
Fichier principal
IWCS-AKLR-final3.pdf (448.73 Ko)
Télécharger le fichier
IWCS-6.ppt (345.5 Ko)
Télécharger le fichier
Origin : Files produced by the author(s)
Format : Other
Loading...