Multi-Source Assessment of State Stability

The wave of revolutions in the Arab world, commonly referred to as the Arab Spring, took the world by surprise. Despite the rich literature on interstate conflict, state stability, and revolution and regime change, the Arab Spring could not be predicted nor fully accounted for by the existing theoretical traditions in the social sciences.

SC Faculty and Researchers

Kathleen M. Carley

Wei Wei

Matthew Benigni

Kenneth Joseph

As protests and demonstrations have broken out in one country after another over the last several years, questions arose as to what mechanisms supported the diffusion of ideas and actions, promoting or inhibiting violence, and eventually enabled successful regime change. New communication technologies and social media were touted as critical to these revolutions. The belief in the power of the Internet was such that in some cases embattled leaders turned off access.

Conventional wisdom still asserts that 'social media help people mobilize and revolt against governments' and argues that Twitter and Facebook directed many of the mass anti-government demonstrations during the Arab Spring and recent embassy events. Yet, the role of the media and the use of such technologies cannot be considered the cause of the Arab Spring or even the recent consulate and embassy attacks. A variety of critical factors all played a role, including but not limited to: environmental change, violation of cultural norms, changing economic conditions, human rights violations, government corruption, dissatisfied youth, increasing food prices, global famine, increased foreign presence, and the persistence of absolute monarchies.

Outcomes varied across countries, each of which followed different patterns, in part due to varied histories with democracy and authoritarian control, alternative respective rhetoric of change, and variations in access to the internet and utilization of social media.

We ask,

  • What role does the cyber-mediated environment play in state-stability?
  • How are traditional and social media used by states and individuals to manage and understand change in this cyber-mediated environment?
  • How can we asses and predict state stability, identify changes in what groups are at risk, and do so at scale given the vast quantity of changing data available on-line through the cyber-environment?

Social media is increasingly becoming a major source of information for populations. However, the grass-roots nature of social media is changing. The majority of news agencies, e.g., BBC, CNN and al-Jazeera use Twitter and Facebook to spread breaking news. Social media is also a major outlet for citizens to express their concerns. In the cyber-mediated environment both social media and traditional media are present and used. Further, the information carried via social media is not completely distinct from traditional media nor more timely. As more organizations and news agencies turn to the use of social media the relative impacts of social media and traditional media on social change become more complex, as does their role in governance.

Despite the increase in attention to the 'cyber world' and a recognition that cyberspace challenges traditional conceptions of influence, diplomacy and security, there is only a minimal ability to track this information and use it to assess or forecast societal level changes, there is a lack of a fundamental understanding of how trust is forged and broken between individuals and institutions in a computer mediated communication environment. The emerging uses of social media technologies, such as Twitter, in contrast to traditional news media, have not been fully examined within the framework of state stability. Empirical analysis is needed to identify what processes are at work in cyber-space and to understand how these processes actually effect the socio-cultural environment and so state-stability.


A mixed-methods, multi-modeling approach is used to support theory development, testing, and model validation. These methods include employing detailed ethnographic analysis, text-analysis in which text-mining using Latent Dirichlet Allocation techniques for topic identification, co-sign analysis for similarity among topics are used, geo-statistics, dynamic network analytics and visual analytics are used for reasoning about the extracted data.

These techniques are used in a progressive and 'stepped' fashion to first identify norms, the lines of balance, critical issues, and indicators of stability, balance, and trust. Then, secondly, we identify groups, topic foci for groups, changes in these indicators and characterize patterns of instability using geo-temporal network and visual analytics. Groups will be identified based on network structure, topic cohesion, and location using community detection technique that we expand to handle both actors and issues. And thirdly, a mixture of statistical approaches is used to characterize behavior, and estimate the likelihood of anomalous change.

Overall, the methodological approach is designed to leverage rich ethnographic description and analysis to provide explanations and interpretations of the statistical, network and visual analytics of the geo-temporally tagged social and knowledge (topic) network data that is extracted from multiple sources. Blending qualitative and quantitative techniques supports the automated coding of media data using text-mining techniques, in-depth analysis of outliers and overall interpretation utilizing culturally informed qualitative ethnographic assessments, and theory testing using traditional and new 'big-data' statistical, network and visual analytics with particular attention to dynamic and incremental metrics for identifying critical actors, issues, messages and groups of interest. These methods are applied to mixed-source data (twitter, blogs, news, trade, geographic information, and archival ethnographic sources) encoded as a series of meta-networks linking people, groups, issues, activities, and location.

Countries: Yemen, Ukraine, UAE, Turkey, Tunisia, Syria, Russia, Qatar. Saudi Arabia, Oman, Morocco, Libya, Lebanon, Kuwait, Jordan, Iraq, Bahrain, Algeria Partial data: Iran, Egypt, Sudan, Palestine

casos logo   minerva logo   onrsmall

We are proud to be working with the following faculty from across Carnegie Mellon:
Huan Liu (Arizona State)
Mia Bloom (Georgia State)

Learn More About This Project

Project Publications

Joseph, K., Carley, K. M., Filonuk, D., Morgan, G. P., & Pfeffer, J. (2014). Arab Spring: from newspaper data to forecasting. Social Network Analysis and Mining, 4(1), 1-17. doi:10.1007/s13278-014-0177-5

Wei Wei, Kenneth Joseph, Huan Liu and Kathleen M. Carley, 2016, “Exploring Characteristics of Suspended Users and Network Stability on Twitter.” Social network analysis and mining.

Kathleen M. Carley, Wei Wei and Kenneth Joseph, Nov 2015, “High Dimensional Network Analytics: Mapping Topic Networks in Twitter Data During the Arab Spring” In Shuguan Cui, Alfred Hero, Zhi-Quan Luo and Jose Moura (eds) Big Data over Networks, Cambridge University Press.

Kenneth Joseph, Wei Wei, Matthew Benigni and Kathleen M. Carley, 2015-forthcoming, “Inferring the Affect of Identities and Behaviors from Text.” Journal of Mathematical Sociology.

Wei Wei and Kathleen M. Carley, 2015, Measuring temporal patterns in dynamic social networks. ACM Transactions on Knowledge Discovery from Data (TKDD).