Data-driven content analysis.
Viacom Inc., parent to companies such as Nickelodeon, MTV and Comedy Central, owns a large library of TV, video, and animated content. Currently, these assets and content lack meaningful metadata making it difficult to perform predictions, content recommendations, and other information retrieval tasks. In particular, content such as animations present unique challenges to analysis and information retrieval. In order to better leverage its content library and better connect with its audience, Viacom needs rich metadata around its content assets.
The Viacom: Natural Language Processing and Machine Learning project aims to extract and visualize themes and topics, including named entities, emotional tone, plot and other characteristics from television content using a large text corpora composed of closed caption files.
The program is led by Yao Wang, Professor of Electrical & Computer Engineering, Biomedical Engineering, and Radiology at NYU Tandon School of Engineering. Yao Wang is a recognized expert in video coding, networked video applications, medical imaging, and pattern recognition.