An Integrated Data-driven and Human-centric Indication and Warning Framework using Broadcast and Social Media

IAI proposes to develop an integrated and focused indicators and warnings (I&W) framework to increase time available for critical decisions and actions, and real-time machine analytics to reduce time to actionable intelligence. In that sense, we will extract and evaluate the fitness/goodness of novel psycholinguistic, graph, and multimedia indicators (and temporal variations thereof) extracted from complementary datasets. We believe that due to the high degree of uncertainties in indicators and large amount reasoning required to interpret indicators in order to take an action that humans are first class citizen in our approach. As such, the features used in detecting indicators will be human understandable (explainable/interpretable), and furthermore that humans can ‘adjust’ these features to determine their sensitivity and effect on the indicator or warning level. The techniques used to transform features into indicators will provide transparency into which features and to what degrees they effect the indicators. Finally, the approach will provide a notion of confidence associated with any reasoning supporting both analytically-driven and open-ended inquiries.