Can cascades be predicted?

On many social networking web sites resharing or reposting functionality allows users to share others' content with their own friends or followers. As content is reshared from user to user, large cascades of reshares can form. Moreover, as a result of the cascade spreading through the network, the network structure may also change by nodes creating new and destroying existing connections in the underlying social network. In this talk, I will discuss a framework for addressing cascade prediction problems. We study large sample of cascades on Facebook as well as Twitter. We find strong performance in predicting whether a cascade will continue to grow in the future. We also discuss ways in which network structure reacts to users posting and sharing content. We find that the dynamics of network structure can be characterized by steady rates of change, interrupted by sudden bursts. Information diffusion in the form of cascades of post re-sharing often creates such sudden bursts of new connections, which significantly change users' local network structure. Our models can successfully predict which cascades will spread far in the network and which ones will lead to bursts in network structure dynamics.

Mining Propagation Data in Social Networks

With the success of online social networks and microblogging platforms such as Facebook, Flickr and Twitter, the phenomenon of influence-driven propagations, has recently attracted the interest of computer scientists, information technologists, and marketing specialists. In this talk we take a data mining perspective and we discuss what (and how) can be learned from a social network and a database of traces of past propagations over the social network. Starting from one of the key problems in this area, i.e. the identification of influential users, by targeting whom certain desirable marketing outcomes can be achieved, we provide an overview of some recent progresses in this area and discuss some open problems.

Information Diffusion Through Adaptive Seeding

In recent years social networking platforms have become an extraordinary channel for spreading and consuming information in real-time. In response, impressive techniques that leverage this infrastructure are being developed, aiming to select users who can spread information effectively. Unfortunately, despite the immense progress made on this problem, state-of-the-art techniques can often become ineffective simply due to the fact that social networks are characterized by heavy-tailed degree distributions, implying that influential users are rare.

In this talk we will discuss a new paradigm called Adaptive Seeding. We will first discuss a remarkable property of social networks which is related to a classic result in the social sciences known as the friendship paradox. We will then describe a framework for designing algorithms that can leverage this structural phenomenon and present some key algorithmic ideas and fundamental challenges. We will discuss both theoretical guarantees and experimental evidence, suggesting dramatic improvements over current-state-of-the-art are achievable.

Tipping Points and Cascades of Opinion Spread in Social Networks

Human behavior is profoundly affected by the influenceability of individuals and their social networks. This talk discusses the dynamics of spread of opinions in such networks using fundamental models for Social Contagion: the binary agreement model (influencing with committed minorities) and threshold model (threshold contact process). In the first one all individuals initially adopt either opinion A or B, and a small fraction of all individuals commits to their opinions. Committed individuals are immune to influence but otherwise follow the prescribed rules for opinion change. We show that the prevailing majority opinion in a population can be rapidly reversed by a small fraction of randomly distributed committed individuals. When committed individuals exist for both opinions, the difference between larger and smaller fractions of them needed for rapid majority conversion decreases as the smaller minority increases. The results are relevant in understanding and influencing the social perceptions of ideas and policies.

We used the threshold model to find efficient spreaders, fast heuristic selection strategies, and impact of clustering on system dynamics. We find that even for arbitrarily high value of threshold, there exists a critical initiator fraction above which cascades become global. Network structure, in particular clustering, plays a significant role in this scenario. Similarly to the case of single-node or single-clique initiators studied previously, we observe that community structure within the network facilitates opinion spread to a larger extent than a homogeneous random network. Finally, we study the efficacy of different initiator selection strategies on the size of the cascade and the cascade window.