Capturing the semantics of individual viewpoints on social signals in interpersonal communication

The Social Web provides a vast source of user-generated content embedding different views on real life events and activities. Although there is a notable effort to utilise this content in personalisation applications, the existing approaches do not exploit the diversity of individual viewpoints for user modelling and adaptation. On the other hand, the semantic web community has developed extensive apparatus to capture and analyse the semantics of content. However, this apparatus has not been used for capturing the semantics of individual viewpoints to empower the next generation of personilisation systems that embrace diversity. This paper introduces ViewS, a framework for capturing viewpoint semantics from textual user-contributed comments on human activities. ViewS has been instantiated for the domain of interpersonal communication focusing on social signals, for which diverse interpretations can be given according to individual backgrounds and experiences. An experimental study has been conducted including participants providing comments on job interview videos. The semantic augmentation of textual comments has been validated with two experts. The results show that semantic enrichment of surface forms can improve the annotation process. Three enrichment methods have been evaluated. The extracted viewpoint semantics has been combined with user profiles to illustrate interesting user patterns for personalisation.