CitationCount for Squacc BiLSTM: a framework for dense video captioning using neural knowledge graph and deep learning