Semantic Content Filtering Using Self-Organizing Neural Networks

COSMOS-7 is an application that can create and filter MPEG-7 semantic content models with regards to objects and events, both spatially and temporally. The results are presented as numerous video segments that are all relevant to the user´s consumption criteria, yet these results are not ranked according to the user´s preferences. Using self organizing networks (SONNs) we rank the segments to the user´s preferences by applying the knowledge gained from similar users´ experience and use content similarity for new segments to derive a relative ranking. To bridge the gap between the user preferences and the content model, an MPEG- 7 model is proposed that uses the hanging basket model to better relate the users ´preferences and usage history to the content.