Multimedia Data-Intensive High-Density Computational Support for Research on Monitoring Cyber Warfar
Abstract
Abstract (Approved for Public Release)Adversarial actors harness social media to conduct disinformation campaigns, provoke hysteria,, radicalize, and coordinate nefarious acts. Social medias vulnerabilities revolve around the lack of effective global policies, enf,orcement, and most importantly, its affordability to develop a false or misleading narrative before facts are released. Narratives, on social media could be weaponized and propagated at frighteningly fast speeds. Such insidious threats that attempt to influence b,eliefs and behaviors must be considered as modern weapons of cyber warfare. Several studies have focused on various aspects of devia,nt activities on social media, e.g., identification of influential actors, their strategies, and tactics. However, these studies foc,us primarily on text-based platforms, e.g., Twitter, Facebook. Rapid growth in popularity of video-based platforms and rampant misin,formation campaigns on them pose a challenge that needs to be addressed. Moreover, due to inadequate computational resources, much o,deo-based social media platforms, YouTube receives more than 300 hours of videos in under 1 minute. Existing research methodologies, need to be adapted to the new information environment. The problem is compounded by the rapid expansion of the social media landsca,pe. Emerging non-mainstream platforms, e.g., tiktok, parler, clubhouse that contain hate speech and disinformation campaign videos g,arner quick but substantial attention. Often disinformation riddled content then spills over to mainstream platforms.For a limited d,ata from YouTube and other video based social media platforms, our ONR-funded and award-winning research has analyzed disinformation, campaigns in the Indo-Pacific region, COVID-19 discourse and others to identify highly sophisticated organizational tactics (e.g.,, deep fakes, shallow fakes, commenter flash mobs) employed b,rms to identify top disseminators, polarizers, and influential coordination structures assisted by trolls and botnets to effect disi,nformation campaigns. However, much of the research is conducted at small scale, as computational analysis of massive multimedia bas,edonline social networks is extremely resource intensive. The requested equipment will provide a high-density computational environm,ent that will help scale up the research efforts by supporting faster and more efficient analysis of multimedia big social data.Th,erequested equipment will help deploy the models and ONR-funded tools (Blogtrackers, TweetTracker, YouTubeTracker) at scale to allow, analysts use the technological framework in gaining tactical and strategic advantage. The equipment will enhance processing, analys,is, and visualization of huge corpus of multimedia online social network data. By supporting a data analytics-bridge between various, tools, the equipment will enable further discovery and analysis, data and knowledge exchange among stakeholders, and interface and, integrate research findings with the Naval Postgraduate School s SmartCloud test and evaluation environment. The equipment will sup,port multidisciplinary collaborations. UALR educates more minority and underrepresented students than any other university in the st,ate. The research supported by the equipment will provide learning opportunities for minority and underrepresented students in STEM, disciplines. Additionally, PI is involved in various undergraduate research training and education programs, which will further the, training and education goals by involving undergraduate students in the interdisciplinary research afforded by the equipment.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 01, 2022
- Source ID
- N000142212318
Entities
People
- Nitin Agarwal
Organizations
- Office of Naval Research
- United States Navy
- University of Arkansas System