DISCRN: A Distributed Storytelling Framework for Intelligence Analysis

Storytelling connects entities (people, locations, organizations) using their observed relationships to establish meaningful stories among them. Extending that, spatio-temporal storytelling incorporates spatial and graph computations to enhance coherence and meaning. These computations become a bottleneck when performed sequentially as massive number of entities make space and time complexity untenable. This paper presents DISCRN, a distributed frame work for performing spatio-temporal storytelling. The framework extracts entities from microblogs and event data, and links those entities to derive stories in a distributed fashion. Performing these operations at scale allows deeper and broader analysis of storylines. This work extends an existing technique based on ConceptGraph and ConceptRank applying them in a distributed key-value pair paradigm. The novel parallelization techniques speed up the generation and filtering of storylines on massive datasets. Experiments with Twitter data and GDELT events show the effectiveness of techniques in DISCRN.