A Meta-Network Approach to Social Influence Campaigns
Kathleen Carley, Larry Richard CarleyCataclysmic changes in how we communicate in cyberspace are dramatically altering our society, allowing information to spread faster, farther, and with less assurance of its accuracy, and enabling groups to form and recruit members on-line to foster social divides. Organized social media manipulation has more than doubled since 2017, with at least 70 nation states employing computation assisted techniques to shape public opinion. Such activity often involves the use of Bots and Cyborgs, used by individual actors or as part of an orchestrated influence campaign by professional marketers. We find that these social influence campaigns are supported not started by Bots and Cyborgs. Moreover, we find that these campaigns can be identified and described in terms of the underlying set of information maneuvers.
Social influence campaigns are routinely conducted in social media by an influencer, or a coordinated team of influencers, sending messages each of which is part of and builds one or more maneuvers. On the surface, the overarching social influence campaigns appear to simply involve new narratives aimed at evoking particular emotional responses. In contrast, we find that these campaigns often involve efforts to shape the online social-networks; i.e., these campaigns are impacting both who is talking to whom and who is talking about what, as well as impacting what ideas or topics are connected to which. Herein a set of identified information maneuvers that take this into account are presented, along with examples of their use, and potential metrics for their identification. Examples of these information maneuvers are drawn from numerous situations – such as elections and disasters, as well as from numerous countries. The overall taxonomy of information maneuvers is referred to as BEND.
Traditional social network metrics and algorithms are useful in this context, such as degree centrality or community detection techniques. However, we find that they often need to be co-applied to both the actual and potential interaction networks, such as who mentions or friends whom, as well as the ideas or topic networks. Further, many of these maneuvers can only be identified through the change in these metrics over time. Finally, we find that the use of subconscious emotional cues that exist within the posts are also indicative of certain types of maneuvers, and even more so for assessing impact. These diverse metrics, these cues, and the way they can be combined to find information maneuvers is discussed. We find that these cues, rather than traditional sentiment analysis, actually provide greater insight into how actors are trying to influence others and the impact of that influence.