Multimodal corpora

Viac o knihe

Empirically based research in multimodality begins with collecting relevant video data for specific research questions. This data is enriched by human coders who annotate the corpus and explore interconnections between modalities such as face, gesture, speech, and posture. Consistency in the data must be validated using quantitative measures, as systematic analysis of observed multimodal behaviors is essential for multimodal corpora. For analysis and modeling, both qualitative and quantitative tools are necessary for browsing, viewing, extraction, and modeling. Additionally, the intercultural dimension is crucial, as multimodal interaction occurs within social contexts. This state-of-the-art survey documents the outcomes of the International Workshop on "Multimodal Corpora: From Models of Natural Interaction to Systems and Applications," held alongside the 6th International Conference for Language Resources and Evaluation (LREC) in Marrakech, Morocco, on May 27, 2008. It showcases current research from diverse fields such as psychology, artificial intelligence, robotics, signal processing, computational linguistics, and human-computer interaction. The volume includes six selected workshop contributions and seven invited research articles, featuring insights from significant international multimodal corpus projects like AMI and SmartWeb, all of which underwent a rigorous review process for significant revisions and extensions b