Last year, the United States Defense Advanced Research Projects Agency (DARPA) funded $42 million of research in the area of strategic communication and social media. The agency noted that current approaches towards military operations and awareness in social media largely involved unsophisticated manual methods, and called for research that would lead to 'systematic automated and semi-automated human operator support to detect, classify, measure, track and influence events in social media at data scale and in a timely fashion.'
It is not surprising that DARPA is funding research into social media: it is now a given that social media environments are important sources of data for understanding the dynamics of the diffusion of information and human behaviour, and there is growing evidence that social media may impact on external events such as civic unrest (the Occupy Wall Street movement and Arab Spring uprisings are good examples).
But DARPA and the Australian defence community probably realise there will be no quick solutions in their quest for improved understanding of social media behaviour and techniques for exerting strategic influence in these new environments.
The challenge stems from the fact that social media sites such as Twitter are generating terrabytes of highly dynamic and semi-structured data each day. While computer scientists are comfortable with working at 'data scale', the behaviour that is being studied is innately social, therefore requiring behavioural models from outside of computer science. On the other hand, although social scientists have been studying social influence for decades in fields such as economics, political science, sociology and marketing science, the conceptual frameworks do not necessarily translate to social media, and the techniques in the social scientist's toolkit are not designed for research at 'data scale'. What's needed is for computer and social scientists to work jointly on these 'big data' issues.
The challenge will be to understand the process of social influence in social media—how people's opinions and attitudes are influenced by their social media interactions—and how this can be distinguished from other social processes such as homophily (the 'birds of a feather flock together' phenomenon) and contextual factors (e.g. mass media).
But it needs to be recognised that while social networks are important in the transmission of information and behaviour in social media, the content or 'stickiness' of the message is also central. And this is where computer scientists (who know about text analysis) and social scientists (who know about social network analysis) can jointly work on computational social scientific approaches to allow defence agencies to better understand the challenges and opportunities posed by social media.
* These are the author’s personal views.