ARS'23 is the ninth of a successful series. It aims at presenting the most relevant results and the most recent methodological developments in Social Network Analysis.
The workshop will be held on May 2-3, 2023, at the "Hermitage Resort & Thermal SPA", located in Via L.do Mazzella 80, Ischia Island, Naples (Italy).
The workshop aims:
For contributed or Poster sessions
|Title and short abstract submission|
|Notification of acceptance|
|Abstract Final Version|
For Young ARS Group
|Title and short abstract submission|
|Notification of acceptance|
|Abstract Final Version|
Young speakers (under 35 years old) are automatically eligible for the youngARS Session and prices.
|Undergraduate/Graduate students||100 €|
Workshop fees cover workshop material, coffee breaks, lunches and social dinner.
Registration and payment of workshop registration fee are compulsory for at least one author of each contributed paper before April 16, 2023.
In this study, we create a new measure of climate change exposure using the Semantic Brand Score (SBS) by analyzing conference calls transcripts. SBS is a measure of brand importance calculated on text data, which combines methods of social network and semantic analysis. Specifically, we first construct climate-related dictionary from Intergovernmental Panel on Climate Change (IPCC) reports using a keyword extraction technique that leverages BERT embeddings. We have then expanded our dictionary with a measure of similarity derived from the analysis of conference calls’ transcripts. Finally, we have measured the importance of climate change topics in each document, using our dictionary, to retrieve a unique measure of climate change exposure at the firm level. We demonstrate the superiority of our model compared to past literature. Additionally, by using the SBS, we can gather information about the sentiment in climate change discourse and the type of climate change exposure. Finally, we conclude our study by highlighting how this measure can be used to estimate the impact of climate policy uncertainty on the equity market. The main contribution of this study addresses a central problem in climate finance finding a firm-level measure to properly quantify risk and opportunities deriving from climate change. Numerous evidence suggest that the effects derived by future regulation and physical events are currently underestimated by market participants, leading to a potential mispricing of asset prices. Providing investors and regulators with a new measure can mitigate the potential negative effect of such mispricing.
Entrepreneurial ecosystems are wealthy environments in which entrepreneurs, firms, and governments can operate frictionless, contributing to innovation and economic growth. The investigation of the structure of such systems is an open issue. We provide insights on this aspect through the formulation of seven network-based principles associating specific network metrics to distinct structural features of entrepreneurial ecosystems. In this way, we aim to support the measurement of the structural characteristics of an entrepreneurial ecosystem and the design of policy interventions in case of unmet properties. The proposed methodology is applied to an original network built on the relationships occurring on Twitter among 612 noteworthy start-ups from seven different European countries. This is a novel way to conceptualize entrepreneurial ecosystems considering online interactions. Thus, this work represents a first attempt to analyze the structure of entrepreneurial ecosystems considering their network architecture to guide policy-making decisions. Our results suggest a partial ecosystem-like nature of the analyzed network, providing evidence about possible policy recommendations.
In growth and development, increasingly sustainable, the role of face-to-face interaction is certainly present, even in an accessory but always usable way. The research, in various academic fields, had investigated the relationship between them connoting it with implicit assumptions such as: the shortest distance, which facilitated face-to-face interaction and led to arranging the exchange of knowledge and the innovation that triggered the local growth, which has often neglected the issue of sustainability. However, the frequency of these interactions has not been considered. Subsequently introduced and considered as a decision variable in the objective function of innovation, which led to underline its importance, also linking it to geographical distance.
A question that helps us in this perspective is: "Exactly what role and importance do face-to-face interactions have in determining innovation?".
To answer this and other related questions, the author proposes a new model that explicitly connects knowledge sharing activities, created by dedicated hubs, and localized innovative activities. Furthermore, among other things, he notes that the frequency of interactions poses challenges for modeling territories because existing models do not take this into account.
Given these assumptions, it is legitimate to ask which questions about social networks, the explanation of these potentialities, could be effective in giving rise to these types of networks which can also be conceived and implemented with due attention to the challenge of sustainability understood in the most evolved forms.
What could be briefly explained could be the field of development through social network analysis which is based on the collection, processing and analysis, methodologically supported, of relational data.
The work to be done could lead to an affirmative answer to the question posed in the title.
This research focuses on remote investment in financial assets (online trading) in Italy, one of the most rising sectors in the Fintech industry. According to recent studies, Italian individual investors are approximately 250.000, while the global population is estimated at around 15 million. Online trading is becoming increasingly popular mostly because of the rapid growth of digital financial services, which has triggered a constant reduction of entry barriers from the demand-side: in fact, to enter the world of online trading no particular requirements are needed, neither in terms of financial capabilities nor in terms of technological skills. At the same time, it is worth noting that online trading is an extremely risky activity, so much that, according with the literature, about 90% of amateur investors loose money consistently. The main objective of this study is to understand how and why ordinary persons enter into the world of “do-it-yourself” finance, investing their personal savings and exposing themselves to a considerable risk of financial failure. To this end, we developed a survey-based analysis (n=540) by which we investigated individual investors’ motivations, operational profiles, attitudes and socio-economic backgrounds. Our survey also included a position-generator questionnaire, which allowed us to map traders’ personal networks and their social capital endowments. In line with the existing literature, preliminary results show that a considerable part of the amateur investors trade because they need additional resources to address family needs. Moreover, most of them perceive online stock trading as a sort of “social lift” that might compensate lacks in their personal networks (which in Italy are paramount for entering the labour market), providing them a concrete - but risky - opportunity of a quick advancement in their socio-economic status.
The present study focuses on the online debate surrounding the Brexit process, five and a half years after the historic referendum that sanctioned the United Kingdom's exit from the European Union. The contentious and polarising nature of the debate, with passionate arguments and little room for compromise or consensus, make it an intriguing case for examination. In our analysis, to explore the evolution of the debate, we employ graph and network theory to analyse a corpus of over 33 million Brexit-related tweets exchanged on Twitter between December 31, 2019, to February 9, 2020.
In this context, graph theory provides a powerful analytical framework for discovering patterns of association between words and topics, which may then be further examined to uncover clusters of often co-occurring terms indicating the most important themes. More specifically, in our work, we use cluster analysis of co-occurrence word networks to investigate the corpus of the Brexit debate on Twitter. The necessity to quantify network similarities poses considerable theoretical challenges. To address these issues and to effectively deal with the networks' geometrical constrains, we adopt a recently established framework for statistical analysis of manifold data.
Overall, this research demonstrates the power of modern analytical methods for analysing social media data and understanding public conversation. The findings may provide fresh insights into the communication and discourse patterns that underpin political decision-making, with consequences for policymakers and researchers alike.
While the online behaviour of organised hate groups has been extensively studied, only recently has attention focused on the behaviour of individuals that produce hate speech on mainstream platforms. One of the most relevant targets of online hating is women. To identify and analyse producers of derogatory, offensive, insulting, denigrating content towards women, we will focus on content shared on Twitter, given the relatively easy access to networked data based on friend/follower relations. We created an extensive corpus of tweets, gathering Twitter contents containing keywords related to a selection of politically-active women, feminists, female entrepreneurs, journalists, and influencers. A misogynistic tailored lexical dictionary was applied to classify the tweets and derive an index of misogyny at the producer level to select a subset of seed users that are more likely to share misogynistic contents. Trainees, who are engaged in an internship program, analysed in detail these seed users, considering their tweets and other metrics such as the number of contents shared and the number of friends/followers. We retrieved the friend/follower networks of the selected seed producers, their features, and textual contents to classify the nodes in the network and to identify which nodes are more likely to share misogynistic contents. The network relational information provided by the intrinsic nature of social network platforms can be included in the classification task. The relevance of relational data is explained in terms of homophily: knowing that members of a particular community are prone to creating abusive content, and knowing that a given author is connected to this community, allows us to use the information beyond linguistic signals. Finally, we investigated the resulting classified network to recover the structure of the misogynistic network.
Project funded by EU Next Generation, MUR-Fondo Promozione e Sviluppo-DM 737/2021
Every year, more and more companies decide to support sporting, musical, and gastronomic events, enabling them to establish long-term relationships or strengthen existing collaborations. In this contribution, we randomly chose three events (Eurovision Song Contest, European Aquatics Championships, and Cous Cous Fest) that took place in three Italian regions in 2022. The existing relationships between thirteen companies that published at least one document of social reporting in 2021 are analysed exploratorily. Specifically, through Network Analysis, the relationships between the companies before and after 2021 are identified thanks to a Google search and then reported in two pre- and post-event adjacency matrices, also noting the relationship types to understand their nature. It is interesting to notice that despite the random choice of events, all the companies are connected. Then, we compare these networks to those deriving from a text-mining analysis with a latent Dirichlet allocation probabilistic topic model applied to social reports. The research is enriched by studying the statistical significance of revealed topics in the social reports. Our findings highlight some main topics (e.g., sustainability, accounting, and finance). The study intends to reflect on the possible association between existing company collaborations and the similarity among their documents of social reporting. Some of them show a high linguistic affinity between companies, which also reflects actual collaborations. Besides, the documents of some companies seem to be more sustainability-oriented, and the relationships initiated in 2022 show that some companies tend to collaborate with those that seem less sustainability-oriented in their social report.
The present research work is framed within the theoretical framework of Science and Technology Studies. In general, the aim is to analyze how science can influence and solve social problems, and in particular, how scientific communication during the Covid-19 health emergency has impacted on public opinion and been politicized.
The study is built starting from the analysis of a substantial dataset of online newspaper articles related to Covid-19, collected from the websites of such newspapers and from Facebook during the early months of the pandemic. The available material was processed using computational techniques. To analyze the communicative and narrative patterns of the media, as well as the way they influenced public opinion and collective behaviors, multilayer semantic networks are used.
Introduction. This paper is part of the field of peacebuilding studies, which is defined as "a constructive engagement that addresses the immediate impact and root causes of episodic and structural violence" (Lederach, 1996; Varker et al., 2015), and is composed of two fundamental aspects: 1. capacity for non-violent conflict transformation; and 2. foundations for sustainable peace and development (United Nations, 2010), with a focus on the second aspect (Taylor et al., 2022). (Taylor et al., 2022).
The Peace, Security, and Inclusion (PSI) triangle is not currently part of institutional educational routes, therefore students' interest in and participation with this issue may remain dormant due to a lack of support.
Objective. This note seeks to determine whether inclusion and gender are among the predominant and recurring themes in the group of students who participated in the "Blackboard for Peace" experiment (Rocca, Zavarrone, 2022) and, if so, to identify the demosocial characteristics of students who share the same vision in terms of PSI to propose the future construction of a student network and indicators based on textual network measures.
The topics that came out of the Rocca-Zavarrone experiment in 2022 and administrative student data on enrolment from the IULM University Student Secretariat were used for this study's research.
Methods. In terms of text analytics and social network analysis, the data matrix is presented as an affiliation matrix (Everett et al., 1997; Borgatti et al., 2013). (Everett et al., 1997; Borgatti et al., 2013). Community identification algorithms will be applied to locate structured affine (= homophilous) groups, with an emphasis on algorithms capable of capturing the dynamics of leaders and followers in natural networks (Borgatti et al., 2022) via basic structures recognized as bicliques.
Outcomes. The construction of peacebuilding networks based on ideals of inclusivity, gender equality, safety, and the establishment of indicators.
Uncertainty has increasingly become an inherent feature of contemporary societies, especially after the Great Recession started in 2007 which undermined the stability of most national economies.
Such economic crisis not only directly affects many families and individuals by lowering their income but also indirectly, by engendering general perceptions of uncertainty about the future. This implies that narratives of the economy affect individual decision-making processes regardless of real structural economic constraints.
This paper aims to understand how media have framed economic issues over time, also examining the narratives of different newspapers in terms of topics and semantics.
To address these research questions, we use data from the LexisNexis database (a unique news archive containing almost two billion articles from the European press over the last 20 years). From the half of 2008 to the end of 2019, through a supervised machine learning approach, we select articles related to economic issues from the ten most important Italian newspapers.
We intend to delve qualitatively into media framing dynamics. For this purpose, we first apply LDA topic modelling techniques to detect different clusters of words and similar expressions that best characterize a specific set of articles, in order to grasp the themes and subjects most associated with the economy in the press. Then, through the pattern of topics co-occurrences, we build a dynamic network that allows us to understand how narrative structures evolve over time. Finally, we compare the narrative structures of different newspapers, based on visual clustering and graph analytical tools (e.g. modularity analysis, community detection), over time.
Climate change is now a topic that involves the public opinion, the international scientific community, and it is an important issue globally. Among its consequences, the impact on the displacement of human populations is of strong interest to researchers and policymakers.
The concept of climate migration itself and the subsequent recognition of the status of climate refugee presents a loosely defined border. The reference scientific literature is filled with case studies, which represent the main useful tool for understanding the phenomenon.
This study starts from the consideration that climate change, social inequality, and adaptive migration are dynamic and interconnected processes. Through a meta-analysis case study, supported by bibliometric analysis, the framework of international reference literature is presented, followed by an analysis of semantic networks for each continent to achieve a synthetic reconstruction of the phenomenon's characterizations in different areas of the planet. The research focuses particularly on some sociological concepts: the role of social inequalities and the adaptive strategies put in place within the broader framework of climate migration.
The use of fuzzy cognitive maps (FCMs) is increasingly common in any field in which it is necessary to investigate the relationship between decision makers’ intentions and choices that need to be made under conditions of increasing uncertainty. Our study shows how collective fuzzy maps can be useful to compare the cognitive representations of four small groups that were involved in as many brainstorming practices. Such method was aimed at reconstructing members’ comprehensive viewpoint of a shared problem-solving set- ting. The research did not require neither a preliminary stage of elements elicitation nor a subsequent arrangement of constructs (like in any repertory grid technique for example), but a single step that reproduced reasoning paths and expected consequences of intervention policies on a specific scenario (as it happens in the oval mapping technique). On the one hand, the comparison of the four maps resulted in graph theory standard indices of each fuzzy cognitive map (hence to show the concept contribution in the FCM). This helped us highlight the role of common issues in developing specific territorial policies and show a possible explanatory framework of the current status. On the other hand, with a kind of process tracing technique we were also able to identify various maps’ singularities that allowed for broader thoughts and considerations on ideas/policies that were not convergent in the different groups.
Social media are increasingly becoming a meeting place for society, where information is shared and where public discussions and debates take place. Studying discussions on social media platforms such as Twitter can provide insights into the role played by social media in modern societies, and network analysis provides powerful tools to investigate the structures that comprise online groups and communities.
This work focuses on exploring the community structure that took part in the political conversation during the last Italian electoral campaign: specifically, we worked on samples extracted from ITA-ELECTION-2022, a large-scale dataset of social media posts discussing the 2022 Italian General Election (Pierri et al., 2023).
Based on a subset of approximately 2,500 users identified as active members, we explored the network community structure with a model-based hierarchical clustering, successfully identifying users with prominent positions. We then focused on the dynamics of the communications within and between these communities looking at the retweet and reply graphs, to evaluate evidence of echo-chambers and polarisation.
Current empirical findings suggest that Twitter conversations on political issues rapidly become polarised, with individuals clustering around different influencers.
Organizational contexts are places characterized by a set of relationships that individuals build during their work activities. These relationships and interactions contribute to the creation and development of organizational culture, understood as the set of values, ideas and implicit knowledge that influence and determine the way in which the surrounding environment is perceived. The aim of the present contribution is to interpret the organizational context of a training company, through a cognitive investigation to outline the salient features of the structure and, in particular, of the corporate organizational culture. By redesigning of the experience gained by the members of the organization, in terms of collaborations promoted in the performance of work activities, we intend to analyse explicit values and underlying assumptions. By means of social network analysis perspective we are able to identify, at the micro level, the social actors who play a strategic role in the organization as well as to define, at the macro level, the topological structure underlying formal and informal interactions established in the organization.
Starting from the case study of Virvelle, an organization operating in Campania region, a questionnaire was administered to members of the organization and a semi-structured interview was conducted with two privileged witnesses, who hold managerial positions. The study was focused on the social relations in organizational contexts as well as the role of culture, showing how useful they can be in explaining the behaviour of the actors who make up the network. The main findings highlighted both the presence of key players within the organization and the configuration of formal and informal relationships among the members. A homophile behaviour appears within the different working areas, underlining as the homogeneities and differences in the organizational culture are spread by ownership and the dominant values of different sub-cultures.
Recently, the research on blockmodeling has seen the development on a sprawling literature on networks representing relations between units in two or more time periods (so-called dynamic or temporal networks). Crucially, the literature offers several techniques to blockmodel such networks which vary in important ways (e.g., the definition of the network, modeling of temporal dependency). In this presentation we will present a dynamic blockmodeling analysis dynamic co-authorship network of social scientists in Slovenia based on data from the COBISS database. After providing a brief summary of the data-selection and preparation processes, the focus will be directed on the presentation and comparison of blockmodels obtained using different methods. Amongst the several methods offered in the literature, the stochastic blockmodeling (SBM) for generalised multipartite networks (also known as MBM; Bar-Hen, Barbillon, and Donnet 2022), the SBM for linked networks (Škulj and Žiberna 2022) and that for multilevel networks (Chabert-Liddell et al. 2021), as well Dynamic SBM (Matias and Miele 2017). Mostly, the presentation will focus on the partition produced by the MBM and the SBM for multilevel networks. This choice takes stake of the results of previous simulation studies (Cugmas and Žiberna 2023 [pre-print]), the specific properties of the co-authorship networks under scrutiny and preliminary results. Overall, these approaches seem the most empirically sounds as well as the most noteworthy from a methodological standpoint.
The men’s football transfer market represents a complex phenomenon requiring suitable methods for an in-depth study. Network Analysis may be employed to measure the key elements of the transfer market through network indicators, such as degree centrality, hub and authority scores and betweenness centrality. Furthermore, community detection methods can be proposed to unveil unobservable patterns of the football market, even considering auxiliary variables such as the type of transfer. These methodologies are applied to the players' transfer of the 40 teams participating in the two major categories of the Italian football leagues (Serie A and Serie B). These operations are not confined to the Italian teams but include teams from all over the world. We consider the summer market session of 2022, at the beginning of the season 2022-2023. Local network measures and the comparison of different community detection methods can help to better understand some peculiarities of the Italian football transfer market in terms of different approaches of the teams, based on their market power.
This paper revisits the literature on environmental conflicts in Protected Areas and proposes a theoretical framework that combines the environmental justice approach with an economic and ecosystem services framework. By integrating these perspectives under the Social-Ecological Network approach, the framework aims to improve the understanding of natural resource governance challenges. This paper also suggests a set of indicators to evaluate and track conservation conflicts in Protected Areas. Simultaneously, a questionnaire model has been developed and will be implemented in an online format, making it easy to distribute and administer. The data obtained can be directly beneficial for Protected Areas, offering new perspectives on the drivers of conflicts. The questionnaire also enables a thorough examination of environmental conflicts across various contexts and governance structures, focusing on the influence of social and social-ecological relational dimensions on the emergence of conflicting trade-offs in property rights over ecosystem services. A two-mode network analysis will be implemented considering the network closure and network heterogeneity. Specifically, the two set of nodes, on which ties and attributes will be collected, are social actors and ecosystem services. As main findings, the present research provides a set of indicators and an alternative to sole case-study, to explore and monitor governance issues arising in protected areas. The Social-Ecological Network approach is able to provide a map of social ecological interactions in the relational dimension of environmental conflicts. Thus, this paper aims to demonstrate the importance of the approach to convey the economic and environmental justice frameworks, in order to target the analysis of Protected Areas conflicts and to guide the selection of the most effective indicators to assess and explore such issues. This approach can lead to more economically, socially, and ecologically sustainable and equitable governance of ecosystem services in Protected Areas and more broadly in natural resource management.
School classes are often confronted with subgroup formation. From a signed network perspective, such groups can be characterized by a high level of positive ties among members inside each subgroup and a high degree of negative ties between these subgroups. Such subgroups are partly formed based on nodal characteristics such as ethnicity, gender or religion. However, such subgroups can be reinforced or countered. One crucial question is which micro-level processes drive subgroup formation. In this paper we focus on religion (non-religious, Christian, Muslim) as a basis for subgroup formation. We know surprisingly little about the mechanisms behind the emergence and stability of religious segregated subgroups. We ask the question, which network mechanisms reinforce them, and which ones might counter such segregation. Relying on social identity and (structural) balance theories, this study investigates the co-evolution of friendship and dislike networks in creating ingroups and outgroups along religious lines. To answer this question, we use a stochastic actor-oriented model (SIENA model) to study the evolution of dislike and friendship among 1,204 secondary-school pupils in 5 German schools. Results show that both religious homophily in friendship and religious heterophily in dislike simultaneously form and maintain high connectedness within (group cohesion) and lack of connectedness between (group boundaries) Muslim and non-Muslim groups. However, we also find that some triadic processes involving a combination of positive and negative ties amplify segregation, while others tend to counter such divisions based on religion.
Corruption measurement is challenging due to the complex, latent, and shadow nature of this phenomenon. Recent approaches for its measurement rely on preventative approaches, and the most prominent literature in this field recognizes red flag indicators as very suitable tools for this purpose. These indicators detect situations at risk of corruption by signaling the presence of anomalies/inconsistencies with the aim of soliciting audits by control authorities. Red flags are often considered in public procurement, a sector particularly vulnerable to corruption, because of the economic interests at stake and the complexity of the several relationships being established in the networks of involved actors. In this work, we use data from the Italian database of public contracts, a massive dataset containing detailed information on every public procurement procedure managed in Italy. Using data about 800 thousand tenders published in 2013-2021, we compute a selection of the most relevant red flags for corruption risk in public procurement: single bidding, contract awarded with discretionary criteria, length of advertising and evaluation periods. At the same time, social network analysis indicators can be derived by considering contracting authorities and/or winning companies as nodes, with shared tenders as the edges connecting these nodes. In this context, we use classical centrality measures (such as degree, betweenness and closeness centrality) and more recent and innovative indicators (such as distinctiveness centrality and rotating leadership). This work allows us to characterize corruption risk in public procurement in terms of features of the networks involved in the public procurement process through statistical modeling, such as (multilevel) linear and logistic regression. Preliminary results show that some social network analysis metrics are more effective than others in predicting red flags and have an important impact in explaining their variability, suggesting that specific public procurement risks are associated with specific networks of issuers and winners.
The distribution of the number of acquaintances among members of a society is a relevant feature of its social structure. Furthermore, the number of acquaintances (or “degree”) is used for estimating other societal features, such as the size of hard-to-count subpopulations or social cohesion. To estimate the degree, the Network Scale-Up Method (NSUM) asks survey respondents about the number of people they know with a set of first names for which name statistics are available. For this method to be precise, a set of names needs to be selected for the survey that jointly represent the population on a smaller scale in terms of relevant traits such as gender or age. Finding the optimal set of names is a combinatorial problem for which this paper provides a solution approach. The approach can serve other NSUM users, and can be applied to any population for which name statistics distributed over different categories are available. We empirically show that our approach successfully provides subsets of names replicating the population distribution for six countries with very different name statistics.
The aim of the presentation is to illustrate the contribution that social network analysis can make, and has made, to sociological gender studies. Nearly absent in the thought of classic sociologists, gender came to the foreground with the first and second waves of feminisms, where not only women became an interest subject of scholarly research, but gender differences, inequalities and biases started emerging in nearly all social domains. While sociological theories of gender can be grouped in three main categories—macrostructural theories, microstructural theories, and interactionist theories—less attention, both theoretically and empirically, has been paid to the essential role that social networks have in forming, reproducing and challenging gender inequalities. Gender differences have however been extensively investigated by social network scholars, especially in the areas of socialization, personal networks, organizations and scientific environments. By reviewing the empirical finding of social network studies, we identify the network mechanisms that contribute to the formation and evolution of gendered social networks in pre-scholar and scholar age, the role of foci in segregating personal relationships of men and women and the resources they can access to, and the consequences that these network structures, and the gendered cultural expectations they carry about, have in organizational and academic settings. By looking at how gender shapes network formations, and how these formations then inform gendered outcomes, we aim to complement sociological studies of gender with the network perspective, therefore specifying further the role of microstructural theories in linking interactionist processes and macrostructural outcomes.
We explore the possibilities of statistical modelling of 3-layer socio-cultural data (see Basov and Kholodova, 2022). In particular, we juxtapose two theoretical perspectives on the creation of meaning in joint practice and interaction. On the one hand, symbolic interactionism argues that joint material embeddedness, especially creative practice, enables creation of shared cultural meanings. On the other hand, phenomenology and social constructivism maintain that shared meaning emerges in everyday life interaction. We jointly model effects corresponding to each of the perspectives using the unique multi-source longitudinal dataset on five European creative collectives that share working and living spaces. The data include both within- and cross-layer relations between individuals, objects, and words. Shared meaning is captured (separately for creative and for everyday objects) as individuals’ similarities in the descriptions of the objects both of them used, based on relations between words applied to the same objects by dyads of individuals. These similarity matrices are combined with social ties between individuals into multiplex socio-semantic networks (Basov and Brennecke, 2017). Series of exponential random graph models (ERGM) for multiplex networks examining the effect of social ties on similarities of meanings of everyday and creative materiality are estimated. In the long term, social ties stimulate similar meanings of both creative and everyday physical objects, which is not the case in the absence of social ties. In the short term, social ties are associated with similar meanings of creative objects (the interactionist argument), but not of everyday objects (the phenomenology/constructivism argument). The issues of (1) simultaneously incorporating different types of social ties captured at different time waves in quasi-longitudinal ERGMs, and (2) conducting proper longitudinal modelling of these data are to be discussed.
Estimating exponential random graph models (ERGMs) with missing data on nodal attributes (e.g., missing gender, age, ethnicity, ...) is currently not possible with the ergm() function of the ergm R-package, the most used software to estimate ERGMs. We have developed a new estimation of Bayesian ERGMs, implemented in the bergmM() function in the Bergm R-package, capable of estimating (Bayesian) ERGMs with missing nodal (and missing tie) data. The algorithm also provides, if desired, multiple imputation of the attributes (and tie-variables) to be used for additional analyses.
In this presentation, we will explain the algorithm, its limitations and assumptions, as well as how to use it in practice.
Cooperation is needed when individuals cannot achieve specific outcomes on their own (Giardini & Wittek, 2019). The relationship between cooperation, gossip, and reputation has been extensively documented in experimental conditions (e.g., Sommerfeld et al., 2008; Wu et al., 2016), as well as some of these relationships partially in field studies. Gossip affects reputation (Burt, 2008); and reputation can be gained or lost through gossip (Foster, 2004). As various disciplines have studied some of these relationships, so are the social mechanisms (Hedström & Swedberg, 1998) identified to explain that gossip and reputation sustain cooperation in groups. This paper aims to empirically study the effect of two social mechanisms of gossip on cooperation in three working units of Hungarian organizations. These mechanisms are distinguished both by the presence of some kind of emotional or instrumental tie between the sender and the receiver of a gossip, and also by the underlying reasons that make cooperation desirable. In this paper we will analyze the multiplex network of gossip, frequent working conversations or friendship (for their respective mechanism to be assessed) and desirable cooperation. At the same time, we will study the structural logic of the network of gossip that emerges among colleagues in the working units. To evaluate our hypotheses, we use Exponential Random Graph Models (ERGM) for the gossip and cooperation networks (Lusher et al., 2013; Pattison & Wasserman, 1999). Preliminary results demonstrate the existence of both mechanisms in these three working units. However, when controlling simultaneously for social bonding and social capital, only the latter remains significant.
The composition of self-assembled social groups is believed to reflect the biological imperatives, personal preferences, and strategies of the group members. A handful of statistical models have recently been proposed to explain the composition of social groups and uncover the mechanisms driving their formation. However, it remains unclear whether these models can be used to test theories of group formation and, if so, how they should be specified. One major issue is that many theories related to preferences (or propensities) for particular partners are defined at the individual rather than the group level. That is, preferences are stated for individual traits of partners, not for distributions of traits in a group.
In this work, we apply methods from game theory to inform the choice of statistics in the Exponential Random Partition Model (ERPM), a recently proposed model for partitions of individuals. Building on previous work bridging ERGM and cooperative game theory, we define a coalition formation game in which the coalition values (i.e., the utilities of the groups that individuals join) are functions of the partition statistics in the ERPM. We describe how the model statistics are linked to individual-level mechanisms driving the formation of groups and show how theoretical expectations of individual behavior can be formalized in the ERPM. We discuss how these findings help interpret the model’s parameters and illustrate the proposed method by exploring different specifications of homophily (i.e., the tendency of individuals to form homogeneous groups).
While exponential random graph models (ERGMs) are not necessarily suitable for large or sampled networks, there are instances where we would like to estimate ERGMs for networks that are partially observed or networks that are too large for regular estimation methods. There are a number of approximate likelihood-based approaches for large networks but assessing the accuracy of these approximate estimates arguably requires comparison with "exact" methods. We consider here a few approaches for Bayesian estimation of ERGMs that are exact in that they are simulation consistent in the MCMC sense. In particular we focus on parallelisation of the MCMC scheme for component ERGMs. These are ERGMs for networks with multiple components, such that the network scales not with the number of nodes but with the number of components. We provide an illustration using data on a large set of criminal actors obtained from a criminal intelligence agency
Sustainable development is a compelling need that commits society to adopt tools for implementing and managing interventions across three macro areas: social, economic, and environmental. Within this context, international programs, such as the UN 2030 Agenda, provide a roadmap for achieving sustainability. States must work towards sustainability by integrating all relevant fields, recognizing that each action has positive and negative impacts on other areas.
Therefore, the 2030 Agenda is founded on a complex structure based on the relational mechanisms between the targets, with all these interdependencies represented by synergistic and conflictual meanings. The core elements of the 2030 Agenda are the 17 Sustainable Development Goals (SDGs), 169 targets, and 231 UN-IAEG-SDGs indicators.
Given the hypothesized existence of a relational mechanism between the SDGs and targets, we aim to: i) estimate the network model that underlies the 2030 Agenda, ii) evaluate the synergies and trade-offs between targets, and iii) identify the priorities for interventions.
Considering a n x p matrix, where the n observations are the 27 countries of the EU area and the p variables are the 2030 Agenda targets, with p>>n, and using Graphical LASSO models, this work estimates the latent network structure of the 2030 Agenda targets. The resulting network is analyzed using Social Network Analysis tools, including global and local measures. Furthermore, once the partition of the target is found, they are compared to the target groups specified by the SDGs of the Agenda to observe any regularities or mixes between them.
This study contributes to a relatively unexplored area of research, namely the management of interactions between the SDGs. Assessing the synergies and trade-offs between targets can be highly valuable as it can provide an accurate monitoring system to states, outlining the priorities that must be pursued over time.
The concept of "working in a network" or "networking" has been spreading for years in the business world. Companies have understood that in order to survive and achieve economic success, it is necessary to abandon individualistic mechanisms and open up to systems of interaction between organizations. The transformation of markets and innovation have led companies to develop relationships in which the network system prevails: organizations that develop connections through information and knowledge exchange and collaborations. The objective of this contribution is to analyze how Fondazione Saccone in Campania presents itself as a hub of competences that has created and promoted a system of interrelationships between companies and territorial entities, favoring the exchange of information among actors of different nature and therefore the creation of common knowledge. The inter-organizational network that is created can thus contribute to the development of "human capital" in an increasingly fragile economy subject to the so-called "brain drain". The definition of the Foundation's network aimed first of all at identifying the actors involved, the different relationships established in terms of collaborations in projects, organization of training courses, conducting research on the territory and dissemination of results through the collection of archival data and data taken from social media over the period 2019-2022. The stability and growth of relationships over time highlight the creation of social capital that seems to stimulate and accompany the action of individual business and territorial entities, making them stronger and more competitive on the market. The inter-organizational network created by Fondazione Saccone thus represents a concrete example of how relationships between companies and entities can generate benefits for the development of the entire territory and for the economic and social growth of local communities.
We introduce a new approach that unifies models for multivariate time series with the Latent Position model (LPM) to model networks from count data. The proposed model provides a hierarchical framework with the Poisson processes. Our framework consists of two well-known models: the log-linear vector autoregressive (VAR) model prominent in the literature of multivariate count time series and the Projection model, a popularly known latent variable model. We integrate the Projection model approach into a matrix of autoregressive coefficients of VAR to study the strength of complex interactions among multiple nodes.
Estimation and inferential procedures are performed using the Hamiltonian Monte Carlo procedure. We demonstrate the merits of our model through a simulation study. We also examine empirically the behavior of the model via application to the real datasets. The canonical design of this model provides a clear understanding of the temporal and contemporaneous relationship in multivariate time series data, describes the network’s global topology, identification of hidden patterns, and forecasts the data.
Opinion dynamics models building upon assimilation to others’ opinions (positive influence) as a core mechanism fail to explain opinion polarization – the tendency of a group to fall apart into opposing camps with increasing disagreement. Scholars have proposed negative influence as an additional mechanism: distancing from the opinion of a discrepant source. However, empirical evidence for negative influence is debatable. Two common drawbacks in studies supporting negative influence are (1) lab experiments lack external validity; and (2) model designs disallow disentangling positive influence from negative influence. To address these drawbacks, we employ the Stochastic Actor-Oriented Model (SAOM) to analyze The Arnhem School Study (TASS) data, a longitudinal dataset that tracks students’ social networks and opinions’ evolution. Two new SAOM effects were developed: the $p$-near similarity and $p$-far similarity effect, capturing the influence of others (irrespective of friendship ties) whose opinions are sufficiently similar or dissimilar, respectively, given a threshold $p$. Preliminary results suggest that for the topic of music taste, a model including positive influence from relatively similar others and negative influence from relatively dissimilar others provides a good fit to the data, controlling for positive influence from friends (average similarity effect). We conclude that our approach can successfully distinguish positive and negative influences in adolescents' networks within classes.
Nowadays, network data integration is a demanding problem and still an open challenge, especially when dealing with large datasets. When collecting several data sets and heterogeneous data types on a given phenomenon of interest, the individual analysis of each data set will give only a particular view of such phenomenon. In contrast, integrating all the data will widen and deepen the results giving a more global view of the entire system.
We developed a novel statistical method named INet algorithm, for data integration based on weighted multilayer networks. Under the assumption that the structure underneath the different layers has some similarity that we want to emerge in the integrated network, we generate a “consensus network” through an iterative procedure based on structure comparison, capable of pulling out important information about the phenomenon under study. The procedure tries to preserve common higher-order structures of the original networks in the integrated one, i.e. neighbourhood. Once obtained the consensus network, we compared it with the starting networks extracting “specific networks”, one for each layer, containing peculiar information of the single data type not present in all the others.
We tested our method on simulated networks to analyse the performance of our algorithm and we analyzed virus and vaccine gene co-expression networks to better understand infectious diseases.
In a sunbelt presentation Everett and Borgatti proposed extensions to the EI index and Yule's Q to take account of overlapping group membership. Unfortunately their formulation did not give the required maximum score in the situation they predicted. In this paper we demonstrate why their suggested approach was incorrect and propose an alternative that does not have the same problem. In keeping with their original approach we propose two forms of the measure one based on time spent in activities and one based on skills or interest.
Many researchers have studied the implications of the COVID-19 pandemic and the associated containment measures on demography, society, and the economy, focusing on direct and indirect effects. Using data from two waves of the SHARE Corona Survey carried out in 2020 and 2021, i.e., during the first lockdown and then after one year, we aim to depict the social network characteristics of older people in Italy, focusing on tracing the changes related to the outbreak of the SARS-CoV-2 virus. In these surveys, there is a section dedicated to the personal social network, and we used such information to build individual networks of relationships. Thus, it is possible to assess differences that emerged in elderly social space. Four network types are identified: No alters (isolated person), Kin (familiar social space), Non-Kin (neighborhood/friends/colleagues), and Extensive (regularly seeing family members together with people outside the family circle). Results show that in the second wave most of the elderly recovered from the first wave (when people avoided social contacts also due to social restrictions) and changed network configuration, making it more comprehensive. Most transitions are from No-Alters and Non-Kin types towards a higher number of social relationships. Then, we model the network change at the individual level to test if some personal characteristics (socio-demographic) or pandemic-related behaviours/conditions impact the probability of changing network type. Finally, some relevant “non-intended” behaviours emerged, mainly related to reducing activities involving meeting people, except visiting relatives. Furthermore, the elderly in rural areas are more likely to increase the number of their social relationships, as well as people claiming that they have not been depressed in the last period. In contrast, age has a negative impact on such probability.
Homophily, that is, the tendency of individuals with a relationship to be similar, is a pervasive pattern of social networks. While the general homophily patterns of the US across standard demographics and attributes are well described in the literature, most of the literature on ego networks has used relatively small samples. This implies that some aspects, such as how different demographics interact to generate homophily, how homophily changes for different subgroups, or how segregation varies geographically in the US, remain unexplored. In this work we leverage a sample of more than 15,000 ego networks from a large-scale online survey, the Covid States Project, with viable samples for most US states. As part of this non-probability survey, we asked respondents to provide information on their three closest alters, including their gender, race, age, partisanship, and how far they live. We have longitudinal data for a subset of these respondents, and, for a larger sample of about 25,000 respondents, if each of their alters is vaccinated against COVID-19. This dataset allows examining how the probability of a strong tie increases when individuals have multiple attributes in common, using case-control logistic regressions. We take a close look at homophily levels for attributes such as race for different SES and age subgroups. In addition, we explore how segregation levels for close ties vary across 40 different US states. By doing so, we are able to detect particularly segregated population segments or areas. We also explore if close ties at shorter distances are more or less homophilous than ties at longer distances. Finally, we describe homophily levels by vaccination status, and, using our longitudinal data, point at potential tie dissolution mechanisms due to discordant vaccination decisions.
Understanding the size and distribution of networks has been a key focus in social network analysis. One widely used tool in surveys for collecting information on the size and distribution of networks is the “name generator” (Marsden, 1987). Previous research has shown that respondents report a low number of alters or, in some cases, an absence of them altogether. These findings have led to several interpretations, including the possibility that the lack of discussants is due to interviewer or wording effects. Additionally, it has been suggested that these results imply that individuals may be socially isolated (e.g., McPherson et al., 2016). From a comparative perspective, this study examines the sociodemographic and social factors associated with the likelihood of individuals not reporting discussants. Our research uses data from the Comparative National Elections Project (CNEP), which includes information from 41 surveys conducted across 22 countries. This study focuses on two specific name generators: whom they talk to about important issues, and whether they discuss political matters with them. Our preliminary results show that certain sociodemographic and social interaction factors can reduce the likelihood of individuals not reporting discussants, such as age, being married, and having a job. However, these factors vary across countries and change over time. To shed further light on this issue, this study also focuses on the CNEP surveys conducted in Italy and Chile over different periods (Italy: 1996, 2013; Chile: 1993, 2017, 2021). The composition of the conversation networks in these countries is described in detail, followed by a discussion of the strengths and weaknesses of the name generator. By understanding the potential limitations of this approach, researchers can develop strategies to improve data collection and analysis. Additionally, by acknowledging the possibility of social isolation, scholars can better understand the challenges some individuals face in several contexts.
Scholars of international migration have long studied the central role that personal networks play in the lives of migrants at different stages of migration and incorporation trajectories. While studies of social networks in specific immigrant communities have grown in number in recent years, systematic comparisons of migrants’ personal networks with those of non-migrants are still rare. These comparisons, however, could be key to understanding disadvantages, resources, and inequalities linked to migration. This study offers a comprehensive and systematic comparison of personal networks between three migration status groups in the San Francisco Bay area of California (USA): first-generation migrants, second-generation migrants, and individuals with no migration background. We use uniquely rich, longitudinal personal network data collected in 2015-2018 with a population-representative panel survey. Analyses consider (1) personal relationships in different domains (e.g., family, social companions, practical support providers, etc.); (2) various characteristics of personal ties, including emotional closeness, spatial dispersion, and difficult or demanding relationships; (2) the structure of alter-alter ties; and (3) the dynamics of personal networks over time. We first describe differences between migration groups (overall and net of confounding sociodemographic factors); then we explore explanations for these differences from existing theories. We find that first-generation migrants have, on average, significantly smaller personal networks, more limited access to social support in all domains, and more geographically dispersed ties. There is no evidence, however, that migrants’ networks are characterized by higher prevalence of difficult relationships and strong ties (operationalized as emotionally closer or multiplex relationships). There is also no evidence, across a wide battery of measures, that migrants are characterized by different patterns of participation in organizations, social groups, and foci of interaction. We conclude that observed differences in personal networks cannot be explained by variation in levels or types of participation in foci of sociability between migration groups.
The world-leading Code of Practice (CoP) for Consumer IoT (Internet of Things) cybersecurity published by the UK government in October 2018 has experienced rapid international uptake since (DCMS, 2018). The CoP laid out thirteen basic and cybersecurity guidelines for consumer ‘Internet of Things’ (IoT) devices informed by industry best practices with the explicit aim of establishing “a set of guidelines to ensure that products are secure by design and to make it easier for people to stay secure in a digital world”. An analysis of the scope and dynamics of how this ‘internationalisation’ progressed presents a unique lens for better understanding the uptake-dynamics of digital technical standards. However, this is yet to be studied and a key reason is because the data of interest based on ‘uptake’ instances globally is typically available in online formats such as company websites that are uncaptured by the academic source remit of classical bibliometric analysis.
Thus, the study of digital technical standards internationalisation is inextricably tied to a methodological challenge and this paper overcomes this challenge in two ways. First it presents a unique methodological 5-step fully-automated approach using Natural Language Processing (NLP)-based co-occurrence networks for studying unstructured datasets and is expected to be useful for future research where classical bibliometrics would not work, as is the case here. Furthermore, this methodology has considerable implications for knowledge discovery processes especially enabling faster data collection by eliminating labour intensive manual data collection and structuring. Second, the findings from this provides a first of its kind (fully automated and randomised) overview of the ways in which the CoP has taken a leading role in shaping global discourse where securing consumer IoT is concerned. This is expected to shed light on the global dynamics that underlie internalisation of digital technical standards.
When critical national infrastructure like hospitals is under cyber-attack, it threatens safety and wellbeing of individuals as demonstrated by WannaCry. To understand vulnerabilities of the healthcare sector and how the sector can take steps to prevent attack, it is important to understand how cyber threats have evolved by looking at the nature of attacks. Through the analysis of news articles, attacks that have become more prevalent requiring action plans can be uncovered. News analysis also helps in understanding the extent to which cyber attacks are localised. Using GDELT (Global Data on Event, Location and Tone), this paper aims to create a keyword co-occurrence network and through a combination of Natural Language Processing and Machine Learning map the temporal evolution of global cyber-attacks. This paper thus creates knowledge maps for uncovering meaningful knowledge components to improve comprehension of cyber attack evolution based on patterns between keywords that emerge in the news. Combining this data with hospital attributes and location, this will be enable deciding on the fist key step in determining what cyber defences would be required and how policy interventions can be developed to help improve the cyber security of Critical National Infrastructures.
2020 has changed how we are all working, including how educational institutions are engaging with everyone within their network. The recent increase in cyber-attacks have had a crippling effect on Higher Education Institutions (HEIs) generating renewed calls for collaborative efforts through sharing Cyber Threat Intelligence (CTI) to generate timely, actionable insights for institutions. In fact, the higher education system is one of complex adaptive systems, where cyber resilience can be supported by social relations that encourages information sharing, cooperation and connectivity among central nodes of the system. Even though researchers have already investigated the cybersecurity collaborative work practices, very little research has emphasised the role of social capital among organisations in the cybersecurity market space. This paper takes a mixed method approach to exploring the structural and relational dimensions of social capital. Social interactions (a form of the structural dimension of social capital), and collaboration (a form of its relational dimension), were significantly associated with the exchange of resources(cyber intelligence information exchange). Our findings show the importance of the network position related to the nodes in bonding and bridging social capital within the higher education sector, which may influence the ability to respond to a cyberattack. These results provide empirical evidence on how social capital that operates through networks propagates throughout cybercommunities.
Given the technical and logistical problems that must be overcome in cyberspace, it is inevitable that many malicious actors operate in (more or less) formal and structured groups. Such groups are social entities, and their interactions and relationships can be the investigate through Network Analysis. Using a well-known and tested dataset, this paper provides a first, empirical analysis of how several malicious hackers groups cooperate and highlights some of their behavioral characteristics. One important conclusion of this study is that malicious actors tend to populate a “small world”, where they consult, coordinate and help each other for their attacks.
In the recent times, firms have seen a significant increase in the number and complexity of cyber attacks. This has led to higher scrutiny in how boards take an active role in dealing with cyber risks and their preparedness to address cybersecurity risks. As such, the governance structures of the firms now have technology experts and board-level risk committees to manage cybersecurity risks well. Firms are not isolated, they are embedded in complex relationships such as interlocking directors where corporate actors are connected through shared directors. As such, this has raised several probing and pivotal questions relating to cybersecurity and the extent to which interlocking firms are able to exhibit better monitoring of cyber risks. Using data from Orbis of publicly listed firms, this paper evaluates how network position of firms have an impact on the cybersecurity practices of the firm. This paper contributes to the crucial understanding of how different network centrality measures provide different benefits in terms of resource access to firms and hence interlocking directorates should be considered as a determinant for evaluating the cyber readiness of the firm.
Spatio-temporal interactive processes, such as alien species invasions, play a key role in ecology. Existing methods studying such processes often simplify the dynamic structure or the complex interactions of the ecological drivers. In this talk, we show how to use relational event modelling (REM) for analyzing patterns of ecological interaction processes at large spatial scales including time-varying variables that drive these dynamics. REM relies on temporal interaction dynamics, that encode sequences of relational events connecting a sender node to a recipient node at a specific point in time. We apply REM to the spread of alien species around the globe between 1880 and 2005, following accidental or deliberate introductions into geographical regions outside of their native range. In this context, a relational event represents the new occurrence of an alien species given its former distribution. The application of relational event models to the first reported invasions of 4835 established alien species outside of their native ranges from four major taxonomic groups enables us to unravel the main drivers of the dynamics of the spread of invasive alien species. Combining the alien species first records data with other spatiotemporal information enables us to discover which factors have been responsible for the spread of species across the globe. Besides the usual drivers of species invasions, such as trade, land use, and climatic conditions, we also find evidence for species-interconnectedness in alien species spread. Relational event models offer the capacity to account for the temporal sequences of ecological events such as biological invasions and to investigate how relationships between these events and potential drivers change over time.
Social networks are systems of actors tied by relationships. These relationships frequently emerge from and are maintained through relational actions, such as instances of communication or exchange. Often, these relational actions can be measured directly, e.g., when people use electronic communication devices. They can be recorded as relational events, i.e., time-stamped sequences of interactions between pairs of actors.
Dynamic Network Actor Models (DyNAMs) can be applied to make inferences about the social mechanisms in which the dynamics of these events unfold. However, the basic DyNAM framework assumes that social mechanisms operate homogeneously over time. This assumption is often not tenable, for instance, in scenarios where the system as a whole is subject to trend shifting. As an example, reciprocity, as the tendency to reply to received messages, may be more relevant in explaining relational actions at an early stage on a newly established messaging channel than after a consolidation period.
Here, we propose an extension of DyNAMs that allows controlling for time heterogeneity by including time-varying parameters. Changes in the parameters are modeled using a Hidden Markov Model, expressing the coefficients' states and switching dynamics for a predefined number of states. We use Hamiltonian Chain Monte Carlo with the No-U-Turn sampler (NUTS) implemented in Stan to make inferences about the model parameters. Using empirical cases, we illustrate the method and discuss how the flexibility of the proposed approach can represent relational event dynamics as a mixture of DyNAMs.
Advice-seeking typically cuts across organizational boundaries by means of informal connections. By using Stochastic Actor-Oriented Models (SAOM), previous research has tried to identify micro-level mechanisms behind these informal connections. Unfortunately, these models assume perfect network information, do not consider threshold-based critical events, such as simultaneous tie changes, and require agents to perform too cognitively demanding decisions. Indeed, in the context of knowledge-intensive organizations, the shortage of high-skilled professionals could create complex network effects given that many less-skilled professionals would seek advice from a few easily overloaded, selective high-skilled, who are also sensitive to status demotion. To capture these context-specific organizational features, we have elaborated on SAOM with an agent-based model that assumes local information, status-related tie selection and simultaneous re-direction of multiple ties. By fitting our simulated networks to Lazega's advice network used in previous research, we reproduced the same set of macro-level network metrics with a parsimonious model based on more empirically plausible assumptions. Our findings show the advantage of exploring multiple generative paths of network formation with different models.
The last 20 years have seen a substantial reduction of the global incidence of malaria, although this reduction has not happened uniformly, resulting in a disproportionate risk of infection among hard-to-reach populations. Most studies that measure uptake of personal protections look at the relationship between individuals’ characteristics to preventive and treatment behaviours. Less attention has been paid to the endogenous social influence dynamics possibly preventing the diffusion of protections. Understanding the role of social networks is important to shed light on the micro-level dynamic of diffusion. Moreover, testing possible interventions in virtual contexts is key to design efficient policy that could change adoption behaviour. This study aims to understand the impact of social influence on the rate of adoption of complementary mosquito bites preventive measures in three indigenous (tribal) villages in Meghalaya, a rural and remote area in North East of India where malaria is still endemic. We asked each eligible villager to name the people within their village they talk to about health-related matters. For each villager we collected information about their individual characteristics (i.e. gender, age, educational background, occupation, etc.). They also had to indicate if they ever use measures to prevent mosquito bites. In order to study the diffusion dynamics within the observed networks, we built an agent-based model of the observed networks. We then ran computer simulations by assuming various implementations of diffusion mechanisms to generate the best fit of the observed adoption rate. Finally, we plan to simulate possible public policy interventions to identify the most efficient measures to increase the adoption rate.
The employment of mercenaries by Italian states between the 15th and 16th century stands at the inception of a military transformation, away from the medieval organization of military power, feudalism, to the emergence of standing armies. The transition from a free to a mediated market of force is observable in the fact that contracts between mercenaries and states tend to stabilise over time, i.e. there are fewer betrayals and a longer period of service to the state. The content of the contracts underwent a change: for instance, the Republic of Venice included some family welfare benefits that made betrayal more costly.
The urge to resort to mercenary companies, the fluctuating patterns of rivalry and expedient partnerships, made contract breachings and sudden changes of sides relatively frequent.
Our hypothesis is that kinship ties played a major role on the dynamic of a contracts between mercenary family-based companies and states.
The dataset was built from the condottieridiventura.it database which collects more than 4000 biographical notes and contractual specifications of condottieri operating in Italy between 1330 and 1600. The dataset is time-stamped: for each mercenary, the dates on which he concluded the contract are indicated and any defections are reported. Direct and indirect family ties were entered for each mercenary. Ultimately, alliances between regional states were reconstructed with temporal dating.
We built a multiplex network composed by two layers. The first one consists of a one-mode network representing kinship ties among condottieri. The second layer consists of a two-mode network representing employment contracts between regional states and condottieri.
Preliminary results are expected to be shown as regards the role of kinship relationships among condottieri on the contract dynamic over time.
The intense economic growth, both commercial and financial, of Castile in the late Middle Ages and at the beginning of the modern period had different causes. One of them concerns the increase in the connection between markets, a process promoted by the growing connection between merchants and financiers at regional, peninsular and European level. This paper aims to provide a first insight into this process of network generation and the increase in the density of relations between financial intermediaries, i.e. money changers or bankers. The nature of their business made it essential to generate solid networks - economic and social - on which to base credit and exchange operations or simply to ensure the correct circulation of information. To achieve this, and working on the principles of the SNA, we intend to show the main relationships established by the local bankers, as well as the different types of network generated according to the economic interest of each banker, and other aspects that made it possible to generate a powerful local bank on which part of the economic growth of the Spanish Golden Age was based.
This paper investigates the private credit market intermediated by notaries in Milan in mid eighteenth century and the first decades of the nineteenth century. The research is based on a dataset of loan contracts drawn up by the city’s leading notaries in a period in which the city was undergoing a remarkable economic modernization. New ventures both in manufacturing and infrastructure required lively flows of capital whereas the massive redemption of public debt by mid eighteenth century provided a vast multitude of individuals and families with hot money.
Compared to institutionalized credit providers (private bankers and later Savings Banks), this ‘informal’ lending market proved its capability to financing innovative ventures with not exclusively land-backed collateralized loans. Thanks to the dominance of a great deal of reputational information, notaries were able to make the private credit market avoid rationing.
Social Network Analysis evidence demonstrates that a pooling equilibrium was averted in this market and a separating equilibrium was reached. Such a market was conducive to the rise of a modern financial-deepened society. A large share of high-middle-sized capitals were employed for financing the most innovative ventures, new partnerships, and the creation of infrastructures, paving the way to economic modernization, while a vast part of minor loans were used for meeting everyday needs, i.e. new consumption, asset management, debt rescheduling.
The citation network of patents citing prior art arises from the legal obligation of patent applicants to properly disclose their inventions. One way to study the relationship between current patents and their antecedents is by analyzing the similarity between the textual elements of patents. Many patent similarity indicators have shown a constant decrease since the mid-70s. Although several explanations have been proposed, more comprehensive analyses of this phenomenon are rare. In this paper, we use a computationally efficient measure of patent similarity scores that leverages state-of-the-art Natural Language Processing tools, to investigate potential drivers of this apparent similarity decrease. This is achieved by modeling patent similarity scores by means of general additive models. We found that non-linear modeling specifications are able to distinguish between distinct, temporally varying drivers of the patent similarity levels that explain more variation in the data compared to previous methods. Moreover, the model reveals an underlying trend in similarity scores that is fundamentally different from those that have been recently illustrated.
In this presentation, we introduce a new class of latent space models to analyze the import/export trade data between a number of European countries. We assume that the probability of having a commercial relationship between two countries often depends on some unobservable (or not easy-to-measure) factors, like socio-economical conditions, political views, level of the infrastructures. To conduct inference on this type of data, we introduce a novel class of latent variable models for multiview networks, where a multivariate latent Gaussian variable determines the probabilistic behavior of the edges. We label our model the Graph Generalized Linear Latent Variable Model (GGLLVM) and we base our inference on the maximization of the Laplace-approximated likelihood. We call the resulting M-estimator the Graph Laplace-Approximated Maximum Likelihood Estimator (GLAMLE) and we study its statistical properties. Using simulations and the real data application, we demonstrate that our novel approach can be very computationally advantageous and that it can well capture many features of interest from the network.
In recent years, hypergraphs have emerged as a powerful tool to study systems with multi-body interactions which cannot be trivially reduced to pairs. While highly structured benchmark models have proved fundamental for the standardized evaluation of algorithms and the statistical study of real-world networked data, these are scarcely available in the context of hypergraphs.
In this talk, we present a flexible and efficient framework for the generation of hypergraphs with many nodes and large hyperedges, which allows specifying general community structures and tune different local statistics. In particular, our model allows to sample synthetic hypergraphs with a variety of desired features, among which we list assortative or disassortative communities, and mixed or hard community assignments. Furthermore, we show how it is possible to condition on additional requirements, such as average degree, degree sequence, and hyperedge size sequence.
The availability of a highly structured benchmark model allows studies that were previously not feasible in the context of hypergraphs. Firstly, it allows comparing community detection algorithms against data generated with known ground truth. Secondly, by conditioning on various structural characteristics, it allows generating hypergraphs similar to real-world data. The samples produced via our benchmark can be then utilized for the replication, statistical study and validation of real data, as well as for statistical hypothesis testing.
To summarize, our work constitutes the first highly-structured, community-based benchmark model for the creation of synthetic higher-order data, and constitutes a substantial advancement in the statistical modeling of hypergraphs.
Presentation based on the following work:
A Principled, Flexible and Efficient Framework for Hypergraph Benchmarking
Ruggeri N., Battiston F, De Bacco C.
Generalized Inference of Mesoscale Structures in Higher-order Networks
Ruggeri N., Contisciani M., Battiston F, De Bacco C.
GitHub repository: github.com/nickruggeri/Hy-MMSBM
A literature on multivariate functional graph models has emerged in recent years. The graphical representation of conditional dependency among a finite number of random variables is indeed appealing in a variety of applications, such as brain connection studies. We want to investigate a novel extension of this methodology that considers spatially and temporally correlated random functions. A motivating example is the analysis of the semantic network formed by Twitter users. The main purpose of our analysis is to track the evolution of the Brexit debate on Twitter across the UK during a specific time frame. By considering the change in a word's usage over time as a functional realization, the semantic network is then defined as a graphical representation of the conditional dependence among functional variables. Since each tweet considered is localized in both time and space, we shall take into accounts such features to properly define the functional semantic networks. To summarize the richness of information provided by the estimated networks we used different descriptive statistics on graph, which underlie the changes in both time and space of the public debate around Brexit. The main consequence of this work is a novel representation of the links between words in a social network based on their monthly trends. Consequently, we offer a different perspective on a public debate, moving beyond classical semantic networks built from co-occurrences of words in a sentence/tweet.
As the world becomes more complex, interdisciplinary team-based research is required to understand wicked problems such as climate change. Yet despite the extensive focus on the essential scientific competencies to undertake these ambitious research projects, educational institutions overlook the importance of collaboration skills. In this paper, we use social network, quantitative, and qualitative data to evaluate team satisfaction, network integration, and productivity of two National Science Foundation funded teams over a two-year time period. One team performed well and continued to grow and integrate, while also reporting high levels of satisfaction and productivity. However, the other team remained similar in size and level of network integration. Though the leaders of this team who are situated in the core of the network reported high levels of satisfaction, the team members in the periphery were dissatisfied. Overall, the team was less productive than the first. As both teams were led by established scientists, the difference in team satisfaction, network integration, and productivity can be explained by the varying levels of collaboration skills. We argue that effective team science training as well as collaborative leadership structures explain the trajectories of the two teams. Given the importance of collaboration skills in scientific activities, more team science training is necessary to prepare the next generation scholars for interdisciplinary careers.
Organization studies increasingly recognize social networks as a lens to understand the effect of social context on creativity (Perry-Smith & Shalley, 2003; Sosa, 2011). Creativity is associated to innovation and different types of social networks may exist between individuals within an organization. When considering the development of new ideas and creative solutions, advice networks play a relevant role (Li et al., 2018; Lomi et al., 2014), since the combination and recombination of knowledge is supported by informal advice sharing (Aalbers & Dolfsma, 2015).
Through advice networks employees deliver and receive information regarding their work-related tasks; moreover, central positions in the advice network provide individuals with tangible and intangible resources needed for innovation (Cangialosi et al., 2021; Gulati & Srivastava, 2014). According to Cangialosi et al. (2021), there are two main reasons explaining the importance – in terms of creativity and thus innovative performance – of centrally positioned individuals in the advice network: first, a central position exposes employees to a wider array of professional information that can be combined to generate and implement new ideas; second, central individuals are likely to be seen as having higher status, which leads to an increase in support from their colleagues and supervisors. Ambidexterity represents ‘successful management of both exploration (e.g., creating new products) and exploitation (e.g., production and implementation of products)’ (Anderson et al., 2014. However, there is still a lack of studies focusing of its association with individual-level centrality in advice networks towards novel ideas.
Therefore, this study addresses this gap by investigating the association between ambidexterity and network centrality – at individual level – when considering idea generation and implementation within an organization. For our empirical analysis, we use original data collected in 2021 from a consultancy company offering services in the field of environmental analysis and health and food safety.
One of the major issues of COVID-19 crisis on maritime shipping is the impact on the port and terminal network evolution and development of alliances which can affect the economic growth and financial investments. Some issues such as reduction in seaborn trade, a decrease in ship calls, delay in suppling necessary items such as food, medical instruments, and disruption in shipping activities are reported related to the impacts of COVID-19 on the maritime shipping, however, to the best of knowledge of the authors, there is no or a few
research on the effects of COVID-19 pandemic on the network structure of the ports and on the formation of new partnerships in terminal container operators.
Thus, this research as a novel study aims to recognize the strategic and behavioural reactions and changes of organizational structures of some terminal operators based on the creation of brand-new relationships among the network of terminal operators due to the impacts of COVID-19 crisis.
From a methodological point of view, the study adopts the Social Network Analysis (SNA) and Stochastic Actor-Based Models (SABM). SABMs are models for network dynamics that can represent a wide variety of influences on network change, and
allow to estimate parameters expressing such influences, and test corresponding
hypotheses (Snijders et al. 2010). The case study focuses on container terminal operators in Italy and their networks (De Martino and Giordano, 2021). Through the application of the Stochastic Actor-Based Model, the changes in network structure of container terminal operators will be analysed in two waves (before COVID-19) and 2021 (after COVID-19) in terms of size (number of equity ties), density (propensity to build equity ties) and functional direction of collaboration (homophily and heterophily).
The concept of circular economy has recently gained the attention of scholars, policymakers, and businesspeople, who see it as a novel approach for addressing the economic and environmental issues caused by the traditional linear model of production and consumption. In this vein, the scientific literature suggests that companies can innovate their business models – and therefore creating value – by introducing circularity elements with the support of digital technologies, which can improve efficiency in resource utilisation. However, there is a lack of empirical studies investigating the state of the art of digital technologies utilization for enabling circular economy; in particular, it is unclear how much digitalisation is intertwined with circularity in the strategic agenda of businesses. This research focuses of large companies, because of their leadership in global markets and capacity to influence other stakeholders, to address the above research gaps. By analysing the most recent (2021) corporate sustainability reports of the companies included in the Dow Jones Sustainability Index Europe, we use content analysis to conduct both a semantic and relational analysis of the text and identifying the relationship linking the salient concepts. Our results show that, despite the recent interest towards circular economy in business, the connection between digitalisation and the circular economy is still largely absent in large European companies’ business models. We confirm the existence of an ongoing trend – especially in some industrial sectors – but an organic discussion in the strategic-organizational dimension for designing and implementing circular business models is not emerging.
This paper discusses the application of Social Network Analysis (SNA) to corporate networks in a long-term and historical perspective. Starting with the basic concepts of corporate networks and the main research themes it has addressed in business history, the paper then introduces how historical quantitative archival data can be played with and turned into excel data suitable for the study of social networks by using software such as UCINET. The paper then provides some examples of how this methodology has been used at both the macro and the micro levels: national corporate networks in Argentina, Chile and Italy from 1900 to 2017 and the social club memberships, partnership ties, and interlocking directorates of J.P. Morgan & Co. in the early twentieth century. Finally, the paper discusses new perspectives for the application of SNA in business history, including the study of other networks than those of directors, i.e., shareholders’ networks, networks that are created by joint membership of think-tanks, syndics, policy-planning group, university board, employers’ associations, philanthropic associations.
The aim of this work is to investigate how deep are the historical roots of the business familism in the South of Italy. Within the framework of the social network analysis and exploiting the logic of surname, we define three measures of familism. Descriptive analysis and Difference-in-Differences models show different trends by economic sectors, and provide some empirical evidence of the persistence of familism over the Italian Unification as a structural feature of Southern business system between 1820 and 1900.
The present paper is the first step to investigate the diffusion of rural and agricultural credit in the South of Italy over the 19th century in the form of cooperative and mutual credit institutions.
Through the specific case study of a micro-reality like Procida, using the approach of the Social Network Analysis, we study the relationships that linked the poor island’s credit environment to the financial system of Naples and its province, that in the 1880s testified of an efflorescence of cooperative banks. The study of the business ownership structure and of the corporate boards network of cooperative banks will be essential to analyze the role that these institutions had in the support of the island’s economy.
Particularly, after a first reconstruction of the numerical consistency of cooperative banks in Procida and Naples, their governance, and their mission, we aim at understanding if cooperative banks really operated in a mutualistic and welfare direction in the island. More specifically, we analyze the ability of Procida rural banks to stay away from the dominant role of the Neapolitan financial elites, to respond to their real mission of "being of help for workers and farmers" and to detach themselves from the “banking carnival” that would have overwhelmed the whole Italian banking system at the end of the 19th century.
This paper aims to build an algorithm of network dynamics with decision-making under incomplete information. Accordingly, it tries to identify if a social planner reduces the influence of individual biases, such as confirmation bias or assimilation bias on agents' actions, and solves a coordination problem. The research questions are the following: " Can the social planner increase social welfare, by manipulating the set of possible invitations and annoyances, without directly changing a network structure?", and " What are the main drivers of increasing social-planner utility functions?" "How do the results change if the social planner has incomplete information or wrong priors about the fundamental variable?" For this research, a "Liberal Social Planner" was created; a process through which network members get suggestions depending on its utility function. The results have potential applications for the management of social media platforms by the owners of these platforms. Platforms can develop robots that can help their users be more informed and more satisfied. As we live in a world of virtual connectedness, people seem to obtain more information from online network peers than from experts.
The patent citation network is a complex dynamic system that reflects the diffusion of knowledge and innovation across all fields of technology. Relational event models (REMs) have been used to describe citation networks. The two main issues in analyzing this network are that the number of citation events is huge and its dynamic structure changes over time. In this work, we aim to overcome these challenges by proposing a computational REM extension, called the Deep Relational Event Additive Model (DREAM).
DREAM employs machine learning concepts to capture the relationships between cited and citing patents as events that occur over time. Each predictor in the generative citation model is assumed to have a non-linear behavior, which has been modeled via a B-spline approach. In order to fit such model to a network of approximately 8 million patents and over 100 million citation events, we estimate the model through a stochastic gradient descent approach. This allows real-time, efficient estimation of the DREAM parameters and the identification of the key factors that drive the citation network dynamics. The spline approach can be extended to include complex relationships between predictors through multivariate interaction splines, leading to a more accurate and comprehensive interpretation of the underlying mechanisms. Our analysis has revealed several interesting insights, such as the identification of time windows in which citations are more likely to happen and the relevancy of the increasing number of citations received per patent.
Overall, our results demonstrate the potential of the DREAM in capturing complex dynamics that arise in a large sparse relational event network, maintaining the features and the interpretability for which REMs are most famous.
Recently, companies become aware that “business as usual” is no more an option to continue to grow (Bocken et al, 2016) and they need to innovate their business model toward sustainability (Geissdoerfer et al., 2017) and, scholars have progressively focused on identifying factors that support this process such as for example the capabilities that fuel the transition. The Dynamic Capabilities (DCs) Framework (Teece, Pisano, & Shuen, 1997; Teece, 2018) has gained centrality and, more precisely, a growing number of studies are talking about “DCs for Sustainability” (Wu, He, & Duan, 2013; Inigo, Albareda, & Ritala, 2017; Oranges Cezarino et al., 2019). An analysis of the evolution of the debate, its sources, and current trends, can help to understand the state of the art of the literature and future research.
We applied a bibliometric approach and our analysis – which comprehends 417 scientific documents published during the last twenty years - uses bibliometric techniques to map the cumulative scientific knowledge (Donthu et al., 2021). This kind of analysis has the potential to show how specific disciplines, scientific domains, or research fields are conceptually, intellectually, and socially structured (Cobo et al., 2011) and, thus, it enables to identify knowledge gaps and potential avenues for future research. Our study reveals the existence of six main thematic clusters, identified through the application of the bibliographic coupling technique (Donthu et al., 2020) enriched through Co-Word analysis (Chang, et al., 2015; Donthu, et al., 2021).
We noticed a growing interest in Sustainable DCs with a recent focuses on Circular Economy (Pieroni et al, 2019; Khan et al, 2020), Big Data (Dubey et al., 2019; Bag et al., 2020) and Supply Chain (Beske et al, 2014; Reuter et al. 2010) treated in a perspective of collaboration and information sharing to achieve sustainability results (Kumar et al., 2018).
Scientific collaboration is an essential driver of research progress and innovation in Science.
As evidence of this, collaboration is increasing in all disciplines, and government policies in international exchange programs aim to promote collaboration among researchers.
Collaborations develop through informal mechanisms (e.g., advice, face-to-face contacts, and exchange of personal knowledge) and formal activities (e.g., writing papers and participating in research projects). Scientific collaborations have several advantages concerning productivity but can also have drawbacks. Even with the comprehensive discussion on co-authorship analysis and productivity, the academic literature still needs to improve the analysis of the pros and cons of co-authorship on scientific productivity, especially comparing different research fields and research communities.
In this contribution, we propose a comparison of the co-authorship networks between two different fields of study for Italian scholars, the statistics community (SECS-S (Statistics, composed of 5 distinct subfields) and management community (SECS-P/08 Management), both belonging to the macro-group named “Area 13” (Economics and statistics).
Data about scholars’ scientific production have been retrieved from the Scopus platform in order to build different longitudinal co-authorship networks at subfield level.
We will compare the different co-authorship networks by considering their topology and authors’ position. We will also analyze productivity and co-authorship structures over time in order to identify the differences in scientific collaboration patterns emerging in the fields under analysis. Proper graph measures will be selected to assess the role of the collaboration configurations in the scholars’ productivity frameworks, with the main aim of comparing the two macro-area in that regard.
Furthermore, we build his/her own ego-network for each scholar to analyze the co-authorship dynamics, underlining new publishing behaviors concerning different individual characteristics (e.g., academic position). This analysis is crucial to understand how the authors are related and how the collaboration patterns change across time and between disciplines.
The present research focuses on two different types of participants in students mobility programs. On the one hand, we evaluate the impact of previously established networks on academic performance of Master degree students who spend two years of their studies in two different universities. On the other hand, we check on the importance of existent networks for undergraduate students that move to another university for one semester.
The study demonstrates that in the first case the students use the networks formed in Master programs to acquire assistance in their studies. The network does not tend to be populated, but it's very cohesive. It provides academic support needed to complete successfully the studies and, in some occasions, to start career. However, the networks formed before starting the Master program serve as a source of emotional support in difficult situations students have in their academic and personal life.
In case of Undergraduate studies, the majority of students prefer to mostly rely on networks established prior to start of their mobility program both for academic and emotional support.
Such difference might be explained by length of mobility program, that in the second case potentially does not permit to establish strong connections with groupmates.
Recently, STEM disciplines - which stands for Science, Technology, Engineering, and Mathematics and encompasses a broad range of scientific fields - have been increasingly recognized for their influence on global economic well-being, particularly given the relevance of hot topics such as innovation, digitization, and the ecological transition, which are indirectly related to STEM fields. In this paper, to inspect the inclination and collaboration of the young generation towards STEM disciplines, we identify the most active countries and institutions in terms of STEM student mobility, taking into account the size of the entities. Data were collected from the European Union Open Portal, and a directed and weighted network of STEM student mobility was obtained. Network analytic methods such as hubs and authority algorithms were applied to reveal the structure of student STEM mobilities and identify the most important parts of the network. Special attention was given to gender differences as they are reported as one of the recurring themes that emerge when discussing STEM subjects. The results confirmed the dominance of male STEM students but also showed that this dominance varies widely across European regions, with major differences in the mobility of incoming and outgoing STEM students among Scandinavian and Mediterranean countries. Specific attention should be provided to try to increase the presence of women in STEM fields, including student mobility.
The present contribution aims at introducing an analytic strategy to study complex networks in which sets of different units are linked. Multimode network formalization and data processing are reported by considering real-data on intra-national university mobility. A community detection algorithm is adopted for partitioning filtered bipartite weighted networks describing Italian student mobility, from the provinces of origin to the university of destination in which specific degree programmes are offered. Clusters' solutions confirm the preferential attractiveness routes of Northern universities as well as the dichotomy between scientific and humanistic fields, recalling the South-to-North and the North-to-North student mobility trajectories.
This contribution stems from the general framework of the analysis of student mobility flows, between universities, in a specific geographical context.
We refer to a specific regional level of analysis, considering the case of the Campania region in Italy.
Starting from raw data measuring the occurrence of the students' churn decision in the transition from the first to the second year of the bachelor degree, defined by the choice to change degree course and/or university, we define flow data on the network that take into account both the students' retention capacity by the universities and the direction of the flows between the universities.
The resulting flow data can be interpreted as a measure of the Odds between dropout and retention. This measure of association defines a system of weights placed on the arcs of the local universities' network.
The analysis and representation of these networks will be aided by the use of specific factorial techniques and social network analysis tools.
Many diffusion models consider a static network on which a dynamic influence process unfolds. We propose a novel diffusion model for temporal social networks. The proposed model establishes directed and weighted influence relationships between any pair of nodes based on two antagonistic components: first, the susceptibility to be influenced (or, conversely, the inertia to change the status quo) and, second, the tendency to grow independent from past influence of others. At any point in time, the influence relationships of a node are expressed in a distribution that describes the proportion to which the node is influenced by others or itself. In these distributions, all indirect and time-respecting traces of influence are accumulated by processing time-dependent dyadic interactions and individual independence rates.
We show that the proposed model generalizes the Friedkin-Johnsen model of opinion pooling with stubborn actors. While the model is only an over-parametrization on static networks, it is a proper generalization on dynamic networks. The main reason is that inertia allows for any past influence in the temporal network to have an impact, whereas the same influence dynamics repeat over and over again in static networks.
Dynamic networks arise from a set of players exchanging temporally ordered interactions. The past configurations of the networks may impact the future ones. The relational event model (REM) entails deepening the underlying dynamics that make the system's players engage each other. Yet, an open-research field concerns the evaluation of the goodness of fit (GOF) of this model, especially when it incorporates time-varying and random effects as well.
We consider a smooth mixed-effect REM estimated via case-control sampling and we propose a cumulative martingale-residual-based approach. We may derive various GOF statistics. For instance, if the model is adequate, the receiver-specific normalized sum of martingale residuals can be shown to behave as a standard normal asymptotically.
We present an empirical application which entails explaining sequences of alien species invasions via smooth case-control REM. The first year a species is detected as alien in a region where it was not native represents our time-stamped relational event of interest. To assess the GOF, we stratify martingale residuals by country. With a significance level of 5 %, no misspecification related to the regions is highlighted.
Different stratification strategies may be applied to martingale residuals, to inspect whether the relevant network dynamics features have been adequately incorporated into the model.
The Russo-Ukrainian war has brought to the centre of the debate the strategic role of energy commodities (coal, petroleum and natural gas) and, consequently, of those countries which are pivotal in trading these commodities. In this paper, we build the Commodity World Trade Network where countries are nodes and the bilateral exports properly weighted are the links between two countries.
Our goal is to obtain useful information from the reconstruction of the network to understand not only the evolution of this sector, but above all to identify the main players and as well as those countries that may suffer from a reduction in trade relations.
We analyse its topological evolution between 1995 and 2020, detecting the main hubs and computing different centrality measures. The three markets present hub and spoke and they are strongly incomplete with low density and small diameter in the overall time interval considered. Moreover, in order to evaluate the systemic risk, we remove some nodes and links looking at the achieved effects both in terms of neighbourhood and the aggregate level. Through the spectral clustering we identify communities which could replicate the real geopolitical alliances. Then, we compute the Gini index of the main centrality measures in order to look not only at the degree of inequality of the distributions in each year but also to its evolution. To conclude, centrality measures are used also to recognize main players of the markets in order to understand the economic and geographic reasons for which they play such a central role.
It is widely acknowledged today that music scenes or ‘worlds’ can be analysed as social networks. In most published cases this has meant analysing music worlds as networks of the individuals who participate in them, whether as artists, audience members or ‘support personnel’ (e.g. managers or sound engineers). Such ‘participant networks’ are important and their analysis is often useful and revealing. However, music worlds can also be thought of as networks of events (e.g. gigs and festivals) linked by flows of both participants and the culture and resources those participants bring with them. In this paper I reflect, both theoretically and by way of empirical analysis, on such 'event networks' and what they might teach us. Because face-to-face events occur in particular places at particular times, for example, event networks allow and even demand that we attend to the spatio-temporal structure and dynamics of music worlds. Networks of events have obviously been captured and analysed before, often in the context of two-mode studies, and recent work with both line graphs and hypergraphs is relevant for what I will be discussing. However, very little has been said with respect to the sociological significance of ‘event networks’ and, properly considered, they raise interesting methodological questions for SNA which are not widely discussed (e.g. flows can only move forward in time such that reciprocation of ties is impossible and centrality scores (e.g. in and out degree) are affected by position in temporal order). The paper aims to fill this significant gap.