Representation, information extraction, and summarization for automatic multimedia understanding