Synthesis of collective tag-based opinions in the social web

This paper presents an approach to personalized synthesis of tag-based users’ opinions in a social context. Our approach is based on an enhanced tagging framework, called iTag, where tags are enriched with structure and expressivity and can be addressed to different features of a resource and weighed by relevance. Our main contribution is a synthesis of the collective opinions that is multi-faceted: it shows different points of view on the same resource, rather than averaging the opposite opinions, or choosing the one with the most supporters. If the social tool provides user modeling and trust mechanisms, our synthesis can also be personalized, taking into account both the user’s social network (considering only the opinions of trusted authors) and her user model (considering only the features the user likes). In addition, we propose an innovative visualization modality for iTags, which allows for an at-a-glance impression of all the opinions on a given resource, including significant differences in point of view. We evaluated the iTag framework to test (i) its expressiveness for providing opinions, and (ii) the effectiveness of our synthesis with respect to traditional tag clouds.