Beyond memories: Weaving photos into personal social networks

A picture is worth a thousand words. A personal photo collection may reveal the complex social interactions between the person and his/her social contacts. As digital cameras have become increasingly pervasive, people are empowered to capture many aspects of their daily activities and social interactions with minimal effort. This research explores the hypothesis that social connections can be extracted from knowledge about the people and events in the pictures. This thesis defines the personal social network (PSN) as a model of personal social behavior, and proposes a practical approach to weaving a personal photo collection into a PSN. The resulting model is used to make photo sharing recommendations by spreading activation over the PSN. The experimental results showed that the PSN constructed automatically from photos can achieve good performance, yielding an average precision of 80% with a recall of 75%.