Socio-computational Methods to Detect and Predict Bot Activity in Novel Information Environments

Period of Performance: 07/06/2015 - 05/06/2016

$80K

Phase 1 STTR

Recipient Firm

Intelligent Automation, Inc.
15400 Calhoun Dr, Suite 190
Rockville, MD 20855
Firm POC
Principal Investigator

Research Institution

University of Arkansas
210 Administration Building
Fayetteville, AR 72701
Institution POC

Abstract

Intelligent Automation, Inc. (IAI) proposes to understand social bots behaviors, extract indicators, develop socio-computational models with predictive capabilities to detect bot activity, and implement them in a mature social media analytics software tool. Our approach will use predictive socio-computational models that exploit context, user, friends, temporal, and network features of social media users. Our models will be matured to further understand emerging sociotechnical behaviors for conflict monitoring and social bots activities from organizational and tactical perspectives. We will exploit adaptive machine learning to efficiently refine our models as bot behaviors and social media landscape change over time. By identifying correlation of bot detection and other social media analytics (e.g., influence detection, community detection), we will enhance bot detection by (i) identifying top propaganda disseminators and polarizers, and (ii) extracting influential coordination structures within social bots. The uncertainty and trustworthiness of analytical results will also be computed, and interactive visualization will further help the analysts to drill down and filter. The models and algorithms will then be implemented and integrated with IAIs social media analytics tool that provides advanced analytics capabilities, search, and visualization in a Data Science as a Service (DSaaS) framework.