Indexing Large Online Multimedia Repositories Using Semantic Expansion and Visual Analysis

The proposed framework automatically predicts user tags for online videos from their visual features and associated textual metadata, which is semantically expanded using complementary textual resources.