ICT  (I) Session 2

Abstract: Humans are social animals that love to disseminate ideas and news, as proved by the huge success of social networking websites such as Facebook or Twitter. On the other hand, these platforms have emphasized the dark side of information spreading, that is the diffusion of private facts and rumors in the society. Usually users of these social networks can set a level of privacy, and decide to whom to show their private facts, but they cannot control how their friends will use this information: they could spread it through other social websites, medias or simply with face-to-face communication. The classic Susceptible-Infectious-Recovered (SIR) epidemic model can be adopted for modeling the spread of information in a social network: susceptible individuals do not know the information, then are susceptible to be informed; infectious individuals know and spread the information, while recovered individuals already know the information but do not spread it anymore. A susceptible individual in contact with an infectious one can become infectious with a transmission probability, while an infectious individual naturally recovers from infection with a recovery rate, turning into a recovered individual. We extend this compartmental model in order to represent several kinds of privacy policies, from unsafer to more rigorous: each individual belongs to a class that models the privacy behavior by tuning the transmission probability, the recovery rate and the susceptibility to information, that specifies the interest of the individual on the information. We calculate a privacy score for each individual based on the privacy policies of her neighbors, so as to infer the local robustness to the spread of personal information. We test our model by means of stochastic simulations on synthetic contact networks and on a small partition of the Facebook social network, provided by few hundreds of volunteers that replied to an online survey. Close

Abstract: Digital data is transitioning from data in search of method, to method in search of theory. But, despite a wealth of clearly pertaining theory in the social sciences, it has proven exceedingly difficult to accomplish a fruitful connection: there appears to be an epistemological incompatibility between the computational approaches applied within digital data study and traditional social scientific research. This paper makes two observations about the epistemological nature of digital data. First, digital data is by nature relational and focused on the interaction between individuals, rather than on their individual attributes. This not only permits the study of emergence through computational models - but requires it. Second, digital systems are “arenas of interaction” where the emergence of social practices intermingles with technological change in the platforms that support said social practices. We’re looking at innovation, quite simply, only substantially faster. We interpret this situation through the lens of what Lane and Maxfield (2006) call "ontological uncertainty," which implies limits to the applicability of formal modeling (Andersson et al. 2014.) We find that digital data typically resides at a difficult epistemological crossroad between "mass-dynamics," which is amenable to computational modeling, and ontological uncertainty, which limits the applicability most types of modeling. This paper suggests a complexity epistemology of digital data that is question-driven and methodologically pluralist, using computational modeling tools to explore emergence, and the opportunities of vast new data sets, but does so within an epistemological framing that enables insights that are situated and reflexive. Close

Abstract: Understanding the behavior of users in online systems is of essential importance for sociology, system design, e-commerce, and beyond. Most existing models assume that individuals in diverse systems, ranging from social networks to e-commerce platforms, tend to what is already popular. We propose a statistical time-aware framework to identify the users who differ from the usual behavior by being repeatedly and persistently among the first to collect the items that later become hugely popular. Since these users effectively discover future hits, we refer them as discoverers. We use the proposed framework to demonstrate that discoverers are present in a wide range of real systems. Discoverers are typically not among the most central nodes in the user-user social network, which indicates that users' ability to early collect future hits is essentially unrelated to users' social importance. We show that due to their ability to early identify future hits, discoverers can be used to predict the future success of new items based on the first few received links. Finally, we propose a network model which reproduces the discovery patterns observed in the real data. Data produced by the model are shown to pose a fundamental challenge to classical ranking algorithms on networks which neglect the time of link creation and thus fail to discriminate between discoverers and ordinary users in the data. Our results bring new insights into the quantitative characterization of users' behavior in online systems, and have far-reaching implications for trend prediction and algorithm design. Close

Abstract: Node similarity significantly contributes to the growth of real networks. In this paper, based on the observed epidemic spreading results we apply the node similarity metrics to reconstruct the underlying networks hosting the propagation. We find that the reconstruction accuracy of the similarity metrics is strongly influenced by the infection rate of the spreading process. Moreover, there is a range of infection rate in which the reconstruction accuracy of some similarity metrics drops nearly to zero. To improve the similarity-based reconstruction method, we propose a temporal similarity metric which takes into account the time information of the spreading. The reconstruction results are remarkably improved with the new method. Close

Abstract: Being able to infer the number of people in a specific area is of extreme importance for the avoidance of crowd disasters and to facilitate emergency evacuations. Here, using a football stadium and an airport as case studies, we present evidence of a strong relationship between the number of people in restricted areas and activity recorded by mobile phone providers and the online service Twitter. Our findings suggest that data generated through our interactions with mobile phone networks and the Internet may allow us to gain valuable measurements of the current state of society. This presentation will allow me to further disseminate my work to an audience with broad interests. It will help me establish a personal network of connections in the complex systems scientific community. This may enable collaborations, which would be of great benefit in my career as a young researchers. Attendees of this talk can expect to see the importance of new forms of data both from a scientific point of view, but also their importance for policy makers and stakeholders. The presentation will be readily accessible to a broad audience, thus maximising the dissemination of the research. Reference: Botta F, Moat HS, Preis T. Quantifying crowd size with mobile phone and Twitter data. Royal Society Open Science 2, 150162 (2015) Close

Abstract: Big Data has become the primary source of understanding the structure and dynamics of the society at large scale. The network of social interactions can be considered as a multiplex, where each layer corresponds to one communication channel and the aggregate of all them constitutes the entire social network. However, usually one has information only about one of the channels, which should be considered as a sample of the whole. Here we show by simulations and analytical methods that this sampling may lead to bias. For example, while it is expected that the degree distribution of the whole social network has a maximum at a value larger than one, we get with reasonable assumptions about the sampling process a monotonously decreasing distribution as observed in empirical studies of single channel data. Also we find, that assortativity may occur or get strengthened due to the sampling process. We analyse the far-reaching consequences of our findings. Close