Scalable Tools for Social Media Assessment.
Abstract
Social cyber security is an emerging subdomain of national security that will affect all levels of future warfare, both conventional and unconventional, with strategic consequences. Social cyber security is an emerging scientific area focused on the science to characterize, understand, and forecast cyber-mediated changes in human behavior, social, cultural and political outcomes,and to build the cyber-infrastructure needed for society to persist in its essential character in a cyber-mediated information environment under changing conditions, actual or imminent social cyber-threats. Technology today is enabling both state and non-state actors to manipulate the global marketplace of beliefs and ideas at the speed of algorithms, and this is changing thebattlefield at all levels of war. The technology that enables this includes bots, cyborgs, the talking-head deep fake, AI altered images and memes. Discovery and assessment techniques exist that employ both network analytics, machine learning, and natural language processing (the latter primarily used for sentiment mining). The work in this area, however, has principally been doneon social media posts written in English. This proposal is to conduct the basic research needed to conduct assessment cross culturally, in multiple languages most notably English and Chinese. There are several aspects to this research, first it involves identifying techniques that will operate in multiple languages, such as bot-hunters and meme-hunters and conducting the research necessary to have them operate in Chinese. Second it involves developing new fundamental metrics of the impact of influence campaigns: including, metrics of polarization, trench information warfare, mass hysteria andresiliency. The will be tested and analyzed in scenarios from multiple cultures using Chinese and English data. Third, as an option, it involves developing a semantic/network based approach to localized context based sentiment and stance mining. Techniques to be used include machine learning, high dimensional network analysis, and natural language processing, combined withtesting and assessment using Twitter data. Cross language/culture comparisons will be used to assess utility. Collaboration: The overall project is in collaboration with the NPS CODA lab, and the DSO Info Fusion Lab (Singapore). CMU will serve as the basic research component, developing newfundamental theories, methods, and measures. Scientific Impact: This work is fundamental to cross-cultural assessment of influence campaigns on line in social media. The tests, and data developed for this effort will enablemulti-language technology testing in the social media space. The research on localized sentiment and stance will enable a new way of thinking about sentiment that is less sensitive to syntactic style, jargon, slang or cant.DOD Impact: This work has the potential to provide the DOD with a new scalable approach to assessing influence campaigns that is cross-culturally validated. In addition, the scenarios used for testing and validation are chosen to be of interest to the DOD, particularly as they relate to Chinese activities. Insights about Chinas use of social media and the way the engage in influence campaigns may also have immediate relevance. In addition this work has the potential to facilitate new training modules at NPS of value to the Navy and Marines.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- May 05, 2021
- Source ID
- N000142112229
Entities
People
- Kathleen Carley
Organizations
- Carnegie Mellon University
- Office of Naval Research
- United States Navy