Skip to main content Skip to main navigation

Publication

Proceedings of the 5th International Workshop on Emotion, Social Signals, Sentiment & Linked Open Data

Björn Schuller; Paul Buitelaar; Laurence Devillers; Catherine Pelachaud; Thierry Declerck; Anton Batliner; Paolo Rosso; Seán Gaines (Hrsg.)
International Workshop on Emotion, Social Signals, Sentiment & Linked Open Data (ES³LOD-2014), Emotion, Social Signals, Sentiment & Linked Open Data, located at LREC 2014, May 26-27, Reykjavik, Iceland, European Language Resources Association (ELRA), Paris, 5/2014.

Abstract

The fifth installment of the highly successful series of Corpora for Research on Emotion held at the last LRECs (2006, 2008, 2010, 2012) aims to help further bridging the gap between research on human emotion, social signals and sentiment from speech, text, and further modalities, and low availability of language and multimodal resources and labelled data for learning and testing. As usually rather labels than the actual data are sparse, this year emphasis was put also on efficient community-shared and computer-supported labeling approaches and on cross-corpora experiments. Following LREC 2014's hot topics of Big Data and Linked Open Data in particular also methods for semi-automated and collaborative labelling of large data archives such as by efficient combinations of active learning and crowd sourcing are featured in this edition – in particular also for combined annotations of emotion, social signals, and sentiment. Multi- and cross-corpus studies (transfer learning, standardisation, corpus quality assessment, etc.) were further considered as highly relevant, given their importance in order to test the generalisation power of models.

Projekte