Computational Social Science: Social Contagion, Collective Behaviour, and Networks  (CSS) Session 1

Schedule Top Page

Time and Date: 10:00 - 12:30 on 21st Sep 2016

Room: D - Verwey kamer

Chair: Taha Yasseri

5000 Social influence and opinion polarization on news websites. A field experiment. [abstract]
Abstract: A long tradition of empirical research has demonstrated that voters? political opinions are strongly influenced by their media consumption. Today, however, voters are not only consumers of media content. On the Internet, users produce, adjust, and evaluate media content and contribute to its dissemination. It has been warned that these new forms of social influence may generate feedback processes that intensify users? opinions and give rise to unintended macro-processes, such as opinion polarization. However, these warnings are based on anecdotal evidence and debated, formal models of opinion dynamics. We conducted a field experiment in order to test (i) competing micro-assumptions about social influence and (ii) hypotheses about the macro-dynamics resulting from social-influence. On an American news website, participants read a short, controversial article that contained a voting tool allowing them to rate their opinions about the issue of the article and showing them the opinion ratings of other users. In a first set of experimental treatments, participants saw different distributions of other users? opinion ratings, which allowed us to test alternative assumptions about how users? ratings are influenced by information about other users? views. This data was also used to calibrate a formal model of social influence, which we used to derive macro-predictions about the collective dynamics emerging from the observed social-influence processes. With a second set of experimental treatments, we were able to empirically test these macro-predictions. In a nutshell, we found strong support for social influence. However, the observed forms of social influence generate problematic macro-processes, such as extremization and polarization, only under very limited conditions.
Michael Mäs, Bary Pradelski, and Bernhard Clemm von Hohenberg
5001 Complex Contagion of Campaign Donations [abstract]
Abstract: Money is central in US politics, and most campaign contributions stem from a tiny, wealthy elite. Like other political acts, campaign donations are known to be socially contagious. We study how campaign donations diffuse through a network of more than 50000 elites and examine how connectivity among previous donors reinforces contagion. We find that the diffusion of donations is driven by independent reinforcement contagion: people are more likely to donate when exposed to donors from different social groups than when they are exposed to equally many donors from the same group. Counter-intuitively, being exposed to one side may increase donations to the other side. Although the effect is weak, simultaneous cross-cutting exposure makes donation somewhat less likely. Finally, the independence of donors in the beginning of a campaign predicts the amount of money that is raised throughout a campaign. We theorize that people infer population-wide estimates from their local observations, with elites assessing the viability of candidates, possibly opposing candidates in response to local support. Our findings suggest that theories of complex contagions need refinement and that political campaigns should target multiple communities.?
Vincent Antonio Traag
5002 Does social physics exist? [abstract]
Abstract: In this talk is concerned with the disscontempt that the very idea of social physics causes some in the humanities and social sciences. I will address the historical origins of social physics, as well as that of its opposition to address the following question: has the data revolution merely resurfaced an old debate in the social sciences, or does data technology necessitate a new understanding of the philosophy of social science?
Frederike Kaltheuner
5003 Coupled dynamics of node and link states: A model for language competition [abstract]
Abstract: In this contribution, we focus on the fact that, while the use of a language can be clearly described as a property of the interactions between speakers ---link states---, there are certain features intrinsic to these speakers ---node states--- which have a relevant influence on the language they choose for their communications. In particular, the attitude of a speaker towards a given language ---which determines her willingness to use it--- is affected by individual attributes such as her level of competence in that language, her degree of cultural attachment and affinity with the social group using that language, and the strength of her sense of identity or belonging to that group. For simplicity, we consider that all individual properties affecting language choice can be subsumed under the concept of ``preference''. At the same time, the evolution of the speakers' individual preferences is, in turn, affected by the languages used in their respective social neighborhoods. In this manner, the problem of language competition can be studied from the point of view of the intrinsically coupled evolution of language use and language preference. Ultimately, this change of perspective can be regarded as a shift from a paradigm in which language is considered only as a means of communication to one in which its tight entanglement with culture and identity is also taken into account.?
Adrián Carro, Raúl Toral and Maxi San Miguel
5004 Quantifying crowd size with mobile phone and Twitter data [abstract]
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. However, existing approaches which rely on human analysts counting samples of the crowd can be time-consuming or costly. Here, we investigate whether data on mobile phone usage and usage of the online social media service Twitter can be used to estimate the number of people in a specific area at a given time. 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 Twittter. Figure 1 depicts the time series corresponding to the communication activities inside the football stadium and the number of attendees at the ten matches that took place during the period of analysis, showing a remarkable similarity between them. As well as being of clear practical value for a range of business and policy stakeholders, our findings suggest that data generated through our interactions with mobile phone networks and the Internet may allow us to gain a valuable measurements of the current state of society.
Federico Botta, Helen Susannah Moat and Tobias Preis
5005 The Bass diffusion model on correlated scale-free networks [abstract]
Abstract: The initial inspiration for this work came from some preliminary results of an analysis of inter-?rm innovation networks in the alpine region of South Tyrol. This analysis con?rmed the importance of network connections in the spreading of innovations, as already reported by other studies, and suggested that the network structure should be explicitly inserted into one of the models most widely employed for the description of innovation diffusion, namely the Bass equation. Our experience with the local network structures also pointed to the importance of trickle-up innovation process, which are absent from the traditional Bass model and have been rarely studied in the literature. Actually, a trickle-up process can be only simulated in a model with a network structure, so we saw here a chance to improve the Bass model under the two respects at once. We have successfully integrated the network structure into the equations of the original model and we have studied in particular the total diffusion time and the partial diffusion times (in the link classes) in dependence on the model parameters. This was done separately in the cases of diffusion originating uniformly in the network, or mainly in the hubs, or mainly at the periphery. Further technical improvements have been the explicit construction of correlation matrices and suitable modi?cations of the differential equations in order to allow negative linear terms (representing sti?er hubs) or stochastic terms (representing random generation of innovation). Results come in the form of accurate time diffusion curves and of numerical values assessing the anticipation effect in the hubs. These results show that even for the traditional trickle-down Bass model, the introduction of the network offers several advantages.?
Maria Letizia Bertotti, Johann Brunner & Giovanni Modanese
5006 Quantifying the Impact of Scenic Environments on Wellbeing [abstract]
Abstract: Few people would deny that spending time in areas of beautiful scenery results in a sense of increased wellbeing. Yet, what if scenic environments have an impact on our health??To date, quantifying this relationship has been limited by the impracticality of gathering large-scale data on humans? perception of the environment. However, data generated through our increasing interactions with the Internet allow us to measure human experiences on a scale that was not readily feasible before. Here, we draw on crowdsourced geographic data from the website Scenic-Or-Not (http://scenicornot.datasciencelab.co.uk/) in order to develop a better understanding of how the aesthetics of the environment may impact our health. Scenic-Or-Not allows Internet users to rate the ?scenicness? of photos from all around the United Kingdom. We combine these ratings with geographic data from the 2011 Census for England and Wales capturing respondents? classification of their health. In order to control for socioeconomic characteristics that may be linked with health, we use deprivation data from the 2010 English Indices of Deprivation.
Chanuki Seresinhe, Tobias Preis and Suzy Moat
5007 Skill games versus gambling: from Poker to financial markets. An old debate faced by Statistical Physics. [abstract]
Abstract: A wide number of human activities can be defined as games, in particular when governed by specific rules and leading to the definition of a kind of ranking. The latter can be defined according to several parameters, as the payoff (i.e. prize) gained by individuals, according to factors as the number and the quality of performed actions, the number of received votes, and so on. While for games, like Chess and (Casino) Roulette, the definition of their nature in terms of skill games or gambling, is quite simple, for other games it is really difficult. For instance, classifying the nature of Poker (i.e. as skill game or gambling) seems really hard and it also constitutes a current problem, whose solution has several implications (from laws to healthcare policies). Similar considerations can be done considering Financial Trading. Notably, there are some recent investigations showing that Trading might be seen as a skill game, while others might support the contrary, i.e. that it can interpreted as a form of gambling. Moreover, in the world of financial markets a number of assets are often indicated as more risky than others, like options, derivatives and 'binary options'. Focusing on these two worlds, i.e. Poker (characterized by a number of variants and rules) and Financial Trading (characterized by different assets), in this talk we aim to present with more details this old, but always current and relevant, debate. Notably, we highlight the prominent role that statistical physics might have for facing this problem, i.e. providing a framework for finding a shared solution for classifying the nature of these games.?Remarkably, some results of these investigations show that the intrinsic nature of some games, as Poker, does not depend on their specific rules, but is strongly affected by the human behavior.?
Marco Alberto Javarone
5008 Disentangling interactions in online social systems using multiplex networks [abstract]
Abstract: Nowadays, millions of people interact on a daily basis on online social media like Facebook and Twitter, where they share and discuss information about a wide variety of topics. However, how information spreads among individuals strongly depends on several factors, being the underlying social structure and the different types of social interactions the most crucial ones. Generally, researchers tend to aggregate or disregard some information to reduce the complexity of the data and of their models. Here we will show that this approach is often not suitable to represent and analyze social systems. In fact, we will show that even the most basic network descriptors, such as nodes' centrality and their mesoscale organization in groups or communities, can be very misleading if the underlying network model is not appropriate. By disentangling interactions in Twitter, we will discuss the most recent advances about multiplex analysis and modeling of empirical social systems, from their complex network representation to their dynamics during exceptional events.
Manlio de Domenico
5009 Effects of Network Structure, Competition and Memory Time on Social Spreading Phenomena [abstract]
Abstract: Online social media has greatly affected the way in which we communicate with each other. However, little is known about what fundamental mechanisms drive dynamical information flow in online social systems. Here, we introduce a generative model for online sharing behavior that is analytically tractable and that can reproduce several characteristics of empirical micro-blogging data on hashtag usage, such as (time-dependent) heavy-tailed distributions of meme popularity. The presented framework constitutes a null model for social spreading phenomena that, in contrast to purely empirical studies or simulation-based models, clearly distinguishes the roles of two distinct factors affecting meme popularity: the memory time of users and the connectivity structure of the social network.
James Gleeson, Kevin O'Sullivan, Raquel Álvarez and Yamir Moreno
5010 Nestedness in Communication Networks: From Information Exchange to Topology [abstract]
Abstract: We develop a dynamic network formation model that explains the observed nestedness in email communication networks inside organizations. Utilizing synchronization we enhance K?nig et al. (2014)?s model with dynamic communication patterns. By endogenizing the probability of the removal of agents we propose a theoretical explanation why some agents become more important to a firm?s informal organization than others, despite being ex ante identical. We also propose a theoretical framework for measuring the coherence of internal email communication and the impact of communication patterns on the informal organization structure as agents come and go. In situations with a high agent turnover rate, networks with high hierarchy outperform what we term ?egalitarian? networks (i.e. all agents are of equal degree) for communication efficiency and robustness. In contrast, in situations with a low agent turnover, networks with low hierarchy outperform what we term ?totalitarian? networks for communication efficiency and robustness. We derive a trade-off that accounts for the network communication performance in terms of both measures. Using the example for a consulting firm we show that the model fits real-world email communication networks.
Alexander Grimm and Claudio Tessone
5011 Economic and Financial Networks [abstract]
Abstract: In this talk I will present the network effect in the dynamics of Financial instruments with particular emphasis on the quantitative measurements of bankruptcy impact and risk evaluation.
Guido Caldarelli
5012 Communications Patterns of Individuals with Different Chronotypes [abstract]
Abstract: In computational social science, it is typical to use electronic data collected from many individuals to address population-level or network-level questions. However, there is a lot of heterogeneity in social systems, and each individual is different from others. Recently, there has been increasing interest in this heterogeneity, and efforts have been put on understanding individual differences and time evolution of behaviors of individuals that form the networks. As an example, electronic records of activity have revealed that individuals have distinct daily activity patterns of communication in their egocentric networks, and these patterns tend to persist in time. An important factor in?determining daily activity patterns is the chronotype of individuals, that is, the propensity to sleep at certain hours of the day. Typically, individuals are categorized into three main chronotypes: morning-active, evening-active, and normative. Here, our target is to study the chronotypes of individuals based on smartphone data, focusing on the alternation of periods of activity with periods of inactivity that can be associated with sleep and to find out whether individuals with different chronotypes have different communication patterns. To this end, we use a rich dataset of electronic records collected from over 800 people for more than a year 3 . We apply Nonnegative Matrix Factorization (NMF) to 48 weeks data of digital activity of users (phone screen on/off events), for extracting the dominant daily patterns in the population. Looking at the emerging ?components?, we can see that they very well match with expected activity pattern of different chronotypes. We identify users of three different chronotypes, based on the NMF component which best describes each user?s activity pattern. Upon identifying individuals of different chronotypes, we divide the data from 48 weeks into 4 different periods which approximately correspond to the four seasons. We look at size of social network (derived from communication data) of users of each chronotype and see that there are distinct seasonal variations in communication patterns and size of social networks for each chronotype as well as across different chronotypes.
Talayeh Aledavood, Ilkka Kivimäki, Sune Lehmann Lehmann and Jari Saramäki
5013 Tracking Protests Using Geotagged Flickr Photographs [abstract]
Abstract: Recent years have witnessed waves of protests sweeping across countries and continents, which in some cases resulting in political and governmental change. Much media attention has been focused on the increasing usage of social media to coordinate and provide instantly available reports on these protests. In this talk, I will describe recent research in which we explore whether the data created through such widespread usage of online services may offer a valuable new source for measurements of behaviour during protests. We analyse a large corpus of 25 million geotagged photographs taken and uploaded to Flickr in 2013. For each week and each of the 244 countries and regions, we determine how many photographs were uploaded with the word ?protest? in 34 different languages in the photograph title, description or tag. In order to determine whether there is a link between the number of protest tagged photos and the number of protest outbreaks, we use data from newspaper reports as a proxy for ground truth. For each of the 244 countries and regions, we determine the weekly number of protest related online articles in The Guardian from 2013. We find that higher proportions of protest tagged photos in a given area and week correspond to greater numbers of protest related articles about that area in The Guardian. Our results are in line with the hypothesis that data on photographs uploaded to Flickr may contain signs of protest outbreaks (Alanyali, M., Preis, T. and Moat, H.S., 2016. Tracking protests using geotagged Flickr photographs. PLOS ONE, 11(3), p.e0150466). These findings illustrate the potential value of photographs uploaded to the Internet as a source of global, cheap and rapidly available measurements of human behaviour in the real world.
Merve Alanyali, Tobias Preis and Suzy Moat