Paradigms for the Construction and Annotation of Emotional Corpora for Real-world Human-Computer-Interaction

摘要: A major building block for the construction of reliable statistical classifiers in context affective human-computer interaction is collection training samples that appropriately reflect complex nature desired patterns. This especially this application a non-trivial issue as, even though it easily agreeable emotional patterns should be incorporated future computer operating, by far not clear how realized. There are still open questions such as which types to consider together with their degree helpfulness interactions and more fundamental question on what emotions do actually occur context. In paper we start reviewing existing corpora respective techniques generation contents further try motivate establish approaches enable gather, identify categorize interaction. %Thus believe possible gather valid relevant data material computing community.