Knowledge-aided Interface for Big Data Streams

The proposed work seeks to demonstrate real-time monitoring of broadcast news streams to support a tactical operations center (TOC). Automatically detected essential elements of information (EEI) can be augmented with additional big data streams (news archives and social media) for relevant historical and socio-geographic context. The resulting situational awareness capability would provide tactical advantage to Warfighters operating in unfamiliar environments. A primary focus of this effort is to exploit multimedia audio-visual data containing speech and images, as well as metadata such as geolocation and personal identification. Handling the complexity and scale of multimedia demands specialized algorithms and infrastructure that go beyond traditional text-based processing of data. Automatic speech recognition (ASR) and machine translation (MT) can exceed a human analyst"s cognitive capacity to simultaneously monitor foreign-language sources. FrameNet, a semantic knowledge database, may be used for query expansion to enable more effective search for related stories. Multimedia processing techniques can be used to filter recently posted social media content to determine the locations and identities of key participants. The diverse multimedia sources must be integrated into an intuitive and informative visualization. These are areas of research in which our team has unique expertise and demonstrated success.