Presentation 12. inVideo — A novel big data analytics tool for video data analytics

Video data is a major format of unstructured data, and should be an indispensable area of big data analytics. Existing search engines and data analytics tools such as Google, SAS, SPSS and Hadoop are effective only in analyzing text and image data. Video data analytics are often neglected due to the complexity and challenges in penetrating into videos. In this presentation, we present a novel tool - inVideo for video data analytics. InVideo is able to analyze video content automatically without the need for the initial viewing by human. Using a highly efficient video indexing engine we developed, the system is able to analyze both language and video frames. The index engine identifies both keywords in the audio and objects or individuals in each frame based on reference files, pictures or knowledge. The time-stamped commenting and tagging features make it an effective tool for increasing interactions in online learning and social networking systems. During the presentation, we will demo how the automatic indexing algorithm works in indexing videos. We will also show how the search engine is able to search videos by keywords; by reference files using the Content-Based Image Retrieval (CBIR) algorithm; by knowledge using the knowledge tree we defined; and search videos with different languages. We will also demo how the collaborating filtering works by leveraging user feedback and in improving search engine accuracy. Tagging is another implementation on inVideo system. Videos with identical tags can be linked together or “cropped” based on the preferences. Learning is an integration of interaction. Tagging can turn a non-interactive linear video into an interactive video. This is vital in assessing interactions and outcomes in online learning and systems and MOOCs. We tested the inVideo system in an online learning environment. As longer videos are less likely to be viewed, we broke up the large videos into a serious of 3-5 minutes video clips using i- Video tool. As a result, the interactions in online classes increased seven folds across 24 classes we tested. Experiments also show that inVideo presents an efficient tool in improving interactions in online learning. We plan to expand the experiments in other areas for further studies.