Human Behaviour on the Web: Evolution, Interactions and Exploitation

Abstract

The Web has a fundamental impact on our life, and its usage is quite dynamic and heterogeneous. Moreover, the Web, and in particular Online Social Networks allow people to communicate directly with the public, bypassing filters of traditional medias. Among the others, politicians and companies are exploiting this technologies to widen their influence. In the talk I will show techniques to capture such usage evolution and analyze people interaction on the Internet. This information allows us to understand how users and web services change over time, and how someone can take advantage of these behaviours. There is a large literature about how to evaluate and influence a social network from an analytic point of view [7]. However, it is often not clear if the hypotheses in the mathematical models are valid in real cases and rarely there is enough ground-truth information in large scale experiments. In practice, we observe in the networks heuristic strategies following a trial-and-error approach and emerging behaviours. This is why I am focusing on capturing the human behaviour, directly measured in the present (and past) Web. Thanks to logs of users' traffic, and by active crawling Online Social Networks, I show how to reconstruct users' online activity and to model their behaviour, thanks also to Machine Learning approaches. We deeply understand the evolution of time spent of the Web by the users and the shifting from static pages to the usage of dynamic user-created pages and content in social networks ([4, 6, 9]). The peculiar social networks and other categories usage and evolution can be seen in [1, 4]. Still, considering a short horizon, usage is repetitive and this can exploited for identifying users even when they are not logged (behavioural fingerprints, [8]). Data from human behaviour can be used for extracting and processing social information, sometimes even without the explicit cooperation of the users, to provide new collaborative services. For example, a new service could be the recommendation of hot news that are obtained from aggregated clicks of entire communities (WeBrowse tool, proposed in [3]). Emerging behaviours of the users can also be exploited for expanding someone's influence. A clear example is the recent political debate in Instagram [5] or in WhatsApp [2]. Results suggest that profiles of politicians are able attract markedly different interactions. Moreover, a small group of very active followers can influence a large portion of the network.