Selected topic: One of the current primary research topics of Symeon is the large scale indexing of and knowledge discovery from social multimedia. The massive adoption of media sharing platforms such as Flickr, Instagram, Facebook and YouTube have created huge opportunities for capturing and understanding real-world phenomena, and for powering innovative information retrieval and multimedia-rich applications and services. To this end, it is necessary to devise multimedia indexing and analysis technologies that can cope with the heterogeneity of online content, its huge scale and the prevalence of noisy or low quality content. A promising research direction is the development of compact yet powerful descriptors for near-duplicate search in combination with scalable clustering techniques (e.g. based on community detection on multimedia similarity graphs). In addition, Symeon is interested in multimedia annotation frameworks based on semi-supervised learning that are capable of effectively fusing different features of multimedia content (text-based, visual and social). Integrating the results of the aforementioned approaches with contextual information carried by the multimedia metadata and providing novel information and media exploration interfaces has the potential of disrupting the way we access and search social multimedia content.

Keywords: Social Multimedia, Knowledge Discovery, Multimedia Indexing, Clustering, Multimedia Annotation, Collective Intelligence