The Dynamics Lab at UCD organises a regular Seminar Series and an Occasional Workshop Series. All details of these events are posted here and via various distribution points including the UCD Geary Institute Mailing List. 



UCD Dynamics Lab Predictive Analytics Workshop:  Simulation Techniques for Policy Analysis and Resource Allocation 

Predictive Analytics and Simulation combines advanced analytical techniques and computational capacity to support complex decision making in public organizations. Predictive simulation allows us to go beyond simple patterns, trends and basic data models, to model the interactive and non linear “policy process” with more accuracy. Moreover the greater volume and accessibility of organizational data and including so called 'Big Data', allows us to generate projections with greater accuracy and improve predictive analysis. Identifying threats and opportunities in the data is the first step, then simulation enables solutions to be tried, tested and optimized before policy decisions are made and action is taken. This workshop will explore some current applications and challenges in this new field computational social science. 

Workshop Speakers Philip Cox, Special Investigation Unit, Department of Social Protection (Presentation) ; Georgij Bobashev, UCD Dynamics Lab. RTI International (Presentation); Duncan Cleary, Irish Revenue Commissioners (Presentation); Gillian Golden, UCD Dynamics Lab (Presentation). 

Location/Time  Old Physics Theatre, UCD Newman House, 9.15am-1pm, 11th September 2013



Thijs Velema  National Taiwan University

Friday 28th June  3-4.30 pm in the UCD Geary Institute Seminar Room

Title:  Career sequences involving organizations of different status positions in European professional football


In their career, professional footballers play for several teams at different status levels. Players are either a big fish in a small pond or a small fish in a big pond, as playing for a football club often involves a promotion paradox in terms of a trade-off between the status of a club in the league and the prestige of a player in the team.

This study examines how career sequences of professional footballers involve organizations of different status positions. The sociological literature on careers identifies two answers to this question: early selection and late selection. Early selection emphasizes that talented youngsters move into high status teams at an early age. Some clubs continue to assess the performance of young players in their squad and filter out those judged not good enough. These players move to clubs lower in the status hierarchy as their career progresses. In contrast, late selection argues that footballers gradually progress from low status teams to more prestigious clubs.

This study uses sequence analysis to examine which of these two answers best describes the career sequences of professional footballers in the top seven European leagues. Optimal matching and cluster analysis identify five groups of players. Some footballers better fit the predictions of the early selection model. They are either selected by the high status teams from an early age and retain their position, or started out playing for prestigious teams and were progressively filtered out and moved down to lower status levels. The careers of other players are better described by the late selection model. These footballers played for low status teams early in their career, but were able to move into higher strata as they developed their career. Careers in professional football thus show signs of both the early selection and the late selection model.


Dr Martin Neumann  

12th June 2013 (Note this seminar will take place on Wednesday 12 June at 1pm in the Geary Seminar Room)


An Ontology of Mafia type Organisations


Extortion racket systems (ERS), of which the Mafia is but one example, are powerful and dynamically expanding criminal organisations as well as economic enterprises that cause considerable disruption to the global financial and economic system. The FP 7 project GLODERS, involving several project partners and stakeholder, will develop simulation models to study, monitor, and possibly predict the dynamics of ERSs. This consists of several work-packages. The objective of ontology development is to provide a link between data analysis and modelling. The formal precision of an ontology enables to detect missing links in the data and to organise a process of getting back and forward between data analysis and development of model code. The analysis concentrates on Mafia type organisations (MTO). First and foremost an ontology should identify key actors and relation between them, such as Mafia–Victim relations, corruption and the internal organisation of a Mafia type organisation. Characteristics of MTO in contrast to organised crime are that it controls a territory (may be geographical or along other such as e.g. ethnic boundaries) which falls under its jurisdiction. An MTO asks for extortion money (or other goods) and offers services such as protection in return. This establishes the legitimacy of a (quasi–) political authority over the territory, implying a close relationship with local political elite. The victims may reject the request, accept it reluctantly or accept and co-operate. The purpose of the ontology development is to dissect micro–mechanisms that enable to generate the emergent systemic properties. 


Dr Alberto Caimo  NUI Maynooth

Friday 31st May 3-4.30 pm in the UCD Geary Institute Seminar Room

Title   Bayesian social network modelling: advances + tutorial with Bergm


The Bayesian approaches for exponential random graph models developed by Caimo and Friel (2011, 2013) represent one of the first complete Bayesian inferential frameworks for these social network models. In this talk we will highlight some of the most important features of these approaches and give some insights of future developments. The talk will be complemented by some fine interactive tutorial examples demonstrated through the use of the Bergm software for R.  (Note: You are advised to bring your own laptop). 


Gillian Golden (UCD

3rd May 2013

Title :

Spatially explicit micro simulation techniques for policy analysis and resource allocation in the Irish education system.


Micro level modelling and simulation have come to the fore of many fields of research in recent times, including in the social sciences. This type of analysis is of enormous potential benefit to the public sector as it allows for emergent effects of proposed policy changes to be tested in a synthetic environment prior to implementation, and may provide a greater comprehension of possible issues that arise in a way that traditional top down analysis cannot. I  propose to build a synthetic population that can be used for micro level analysis and simulation in the Irish education system, and to examine a number of policy issues which are of current relevance to see how micro simulation can provide a deeper insight into these issues. The talk will cover a brief overview of relevant literature and results in the area, an outline of my proposed approach for creating the synthetic population, including the types of fitting and selection algorithms under consideration, and an example from the education sector of how the model could be usefully employed for policy analysis.


Eamonn O'Loughlin (UCD

18th April 2013

Building Rich Social Network Data: A Schema to aid in designing, collecting and evaluating social network datasets

Abstract: During the past number of decades the application a Social Network Analysis (SNA) techniques have become significantly more pervasive among sociologists, statisticians and computer scientists, all while the size, scope and complexity of analysed network data has grown. This growth has in part been driven by technological advances (and societies response to those advances) that have resulted in a reduction in the cost associated with collecting and analysing information about social network. However, these technological advancements have increased the difficulty of collecting quality datasets. Compared to multi-dimensional data, there is now a significantly larger amount of methodological and design decisions that must be taken into account when creating a social network dataset, and these decisions must be taken with care, as the features of a dataset determines whether or not it is suitable for particular types of analysis. These design decisions are more than just implementation decisions (e.g. what data structure to use), and are easily be overlooked.

In this paper we propose a standard schema for social network data. A standard schema is a way to define the structure, content, and to some extent, the semantics of a dataset. Our proposed schema defines the most common features that social network datasets may have in a consistent way, allowing for the structure, content and scope of the social network data to be easily assessed and shared. For example, network data may include: node attributes, data for multiple observation periods (dynamic), parallel event data (e.g. vote outcome or committee decision), specific boundary conditions, or known communities. Our schema is designed to capture the underlying informational structure of a dataset and support communication and sharing of social network datasets between researchers irrespective of their discipline or location. After defining the schema, we demonstrate examples using publically available datasets. We then discuss some uses of the schema, which includes: (1) validating a proposed social network data collection design and strategy prior to execution; (2) guiding others (who are holding proprietary or sensitive data) on which features are important to help them create an anonymised, aggregated or a sample dataset without losing the most valuable features.

This work was based upon an analysis of over 150 social network datasets, prepared by the dynamics lab at University College Dublin. This repository of datasets has been made public, and is available on the Dynamics Lab website at


Dr Bülent Özel (Istanbul Bilgi University)

22nd March 2013

A Multi-agent Simulation Model on Individual Cognitive Structures and Collaboration in Sciences

Abstract:  In the first half of the talk, patterns of co-authorship in sciences as of impact of individual cognitive structures on collaborations are discussed. The discussion elaborates upon mutuality of knowledge and social structure theory borrowed from sociology of knowledge literature, where knowledge is perceived as an essentially social and societal category. The findings from an empirical study will be presented. The study covers business management academia in Turkey. It is seen that, academicians publishing at international arena are merely embedded in strongly connected publishing groups. Those locally prominent scientists, however, observed to exhibit distinct cognitive structures compared to their peers, they hold certain part of their knowledge exclusively, thus knowledge-wise they don’t resemble the rest, but they keep a level of common knowledge with the rest of the community. In the second half of the talk, a multi-agent model of collaboration in sciences is presented. The model develops upon empirically observed patterns. It specifically aims to analyze role of cognitive similarities and dissimilarities on agent's incentive to pick a collaborator, as well as, at modeling outcome of collaborations. The presentation concludes discussing various collaboration scenarios and their implications on diffusion of knowledge via collaboration.



Xiong Hang (UCD

8th March 2013

Exit rights, Exit costs, and Self-enforcing Agreements in Cooperative Teams: an ABM Approach


Abstract: This study uses insights from analytical models in order to develop an agent-based model of self-enforcing agreements in cooperative teams. Self-enforcing agreements amongst team members in a cooperative can provide sufficient incentives for members in the absence of third-party monitoring. Granting exit rights and imposing exit costs are two ways of ensuring that self-enforcing agreements are maintained by cooperative members. It is argued that workers’ preference for leisure (i.e. laziness), and the diversity of those preferences, have substantial effects on the effectiveness of penalties. This model sets worker preferences continuous and distribute amongst team member with different statistical distributions (normal distribution and uniform distribution). Statistical analysis of the simulation results indicate: (1) exit costs are necessary for efficient self-enforcing agreements when worker preferences are sufficiently diverse, and (2) the distribution of worker preferences has a statistically significant effect on the effectiveness of penalties. 



Paul Wagner  (UCD

1st March 2013

Modelling Policy networks for managing climate change in Ireland

Despite there being almost unanimous agreement by climate scientists about the dangers resulting from global climate change there has been no similar convergence on an agreement of what should be done in response. This is because the degree to which a society takes the necessary steps to reduce its greenhouse gas emissions does not just depend on having the right institutions in place to address the problem; it also depends on the level of public engagement and social mobilization around the issue. Social mobilization arises when a group or an individual begins to see an issue as a problem and tries to organize others in order to advocate for policies that aim to address it. What to do about climate change is a polarizing question in many societies because it demands a substantial increase in the government’s intervention in the economy, and an increase in the regulation of people’s everyday consumption practices. The government actions that are necessary in order to tackle climate change challenge some strongly held ideological beliefs about the role of government, and would place unwanted demands and restrictions on economic interests. Once policies begin to be debated, stakeholders search out other individuals or groups with whom they share policy preferences so as to coordinate their efforts in order to shape policy outcomes to their joint liking. These coalitions that form in opposition or in favour to government policies comprise of actors from many different sectors of society: from the institutions of the state, the private sector, from civil society, and from non-governmental organizations. The goal of these coalitions is to try and frame the problem of climate change, shape the debate about what the response should be, and to attain their preferred policy outcomes. The goal of this study is to analyse the structure of the Irish climate change policy network and the internal dynamics of the coalitions found therein. A number of hypotheses will be developed about the factors that enable or constrain the emergence of coalitions and how they may be affecting the shape of Ireland’s national climate change policies. 


Bei Gao (UCD)

22nd February 2013

Analyzing the Irish Renewable Energy Innovation Networks

Facing the scarcity of fossil fuels, severely environmental pollution and the climate change caused through combustion of conventional energy, it is widely acknowledged that an inevitable transition of massive launch of renewable energy technologies is taking place. In order to make cleaner energy supply sustainable and affordable, robust control on energy industry by governments is not uncommon. Policy initiatives and government funding provide a heavy incentive for early stage R&D, and collaborative networks between governmental agencies, firms, universities and research institutions. However, the industrial networks tend to dissolve pre-maturely due to the specific characteristics of the innovation target. The aim of this research is to analyze the performance characteristics and development opportunities of innovation networks in renewable energy sector, and to investigate policies designed to promote renewable energy technology adoption and commercialization. It takes the Irish context as a case study, provides a framework to get an insight on relevant data to the energy transition, and visualize the possible outcomes of selected innovation policies with Social Network Analysis tool. The study focuses on three major aspects: First is policy influence on research directions and industrial architecture change (EU/Irish policy documents study, Capacity of energy implemented, No. of new firms established); Second is firms’ innovation performance characteristics according to policy change (Dataset: Community Innovation Survey 2001-2011& Interviews); The third is the development opportunities of collaboration networks and the impact of such network on firm’s performance (Dataset: Community Innovation Survey 2001-2011 & Interviews). Expected outcomes consist of selected scenarios of the reconfiguration of existing industry architectures influenced by innovation policy and interaction of different players. Prominent indicators for economic and social impacts will be identified.


Dr. Harutyun Shahumyan (UCD)

15th February 2013

DL Computational Social Science Seminar Moland model 2013

Application of the MOLAND Model for the Greater Dublin Region

Presentation:In the past few decades, researchers have made considerable progress in improving the capabilities and usefulness of Geographic Information Systems (GIS) for urban management, regional development and planning policy evaluation. However, there is a lack of tools available to compare and contrast policy scenarios, particularly tools which integrate land use, transport and socio-economic variables. The MOLAND model is one such tool which has been developed to support decision-makers working within policy and planning development. It is a Cellular Automata based model sponsored by the European Commission’s Joint Research Centre. The MOLAND model uses GIS as well as socio-economic datasets and runs on a modelling framework Geonamica which allows dynamic integration of a variety of spatial models. The MOLAND has already been applied in over 20 territories in Europe. It has been adapted and calibrated for the Greater Dublin Region by the Urban Environment Project funded by the Environmental Protection Agency Ireland.  Currently, the model is being used to simulate possible scenarios of future settlement patterns in the Greater Dublin Region. Some of the research results have been considered by regional authorities in the recent update of the Dublin Regional Planning Guidelines


Dr. Derek Greene (UCD)

8th February 2013

DL Computational Social Science Seminar Green 2013





Separating News From Noise on Twitter

Twitter introduced "user lists" in late 2009, allowing users to be grouped according to meaningful topics or categories. Lists have since been adopted by journalists and media outlets, such as Storyful, as a way of organising users and content around breaking news stories. Thus the curation of these lists is important – they should contain the key information gatekeepers and present a balanced perspective on a news story. Manual curation of lists is one way to overcome this problem, but is time consuming, and risks incomplete coverage. To support the curation process, we have developed social network analysis techniques that support online journalists in the curation task, processing large volumes of Twitter data to identify key participants. Currently this system is used by Storyful to monitor hundreds of news stories across the globe. We have also addressed the problem of identifying implicit communities of users on Twitter. Here a community may be a group of microblogging users who post content on a coherent topic, or who are associated with a particular viewpoint in relation to a news story. Finally, we examine whether it is possible to curate users and content on a per-community basis to identify different views around a news story, thus avoiding a "filter bubble" effect where the majority view dominates.











Dr Gianluca Miscione (University College Dublin)

16th November 2012

DL Computational Social Science Seminar Miscione 2012

Using Social Network Analysis to profile people based on their e-communication and travel balance

ABSTRACT. The new era of Information and Communication Technologies (ICT) enables people to communicate and interact with each other in new and different manners, changing the way they conduct their daily lives. This change inevitably has significant implications for physical travel in the age of electronic communication (e-communication). This paper aims to provide greater insight into people’s travel behavior based on their e-Communication to Travel balance (C/T balance). This balance represents the ratio between an individual’s e-communication and physical travel. The analysis studies the relevance of social ties in Ahmedabad, India, as a source of explanation of social activity, thus travel, undertaken by individuals. It is hypothesized that the C/T balance emerges from an individual’s social network characteristics. The ability of an individual to engage in social activities not only depends on the individual’s socioeconomic and lifestyle attributes, but also on the ‘modality’ of such interactions (e.g. physical travel or e-communication). The different modalities create different social networks. Each network represents a particular flow of potential activity travel generated by interaction between the individuals. These networks have been clustered on the basis of their C/T balance to get distinct people’s profiles that can be used to target transport and ICT policies better.


Dr Pablo Lucas (University College Dublin )

2nd November 2012

DL Computational Social Science Seminar Lucas 2012

Integrating Collective Decision-Making Models and Agent-Based Simulation

Abstract. Collective Decision-Making Models (CDMM) are mathematically deterministic formulations (i.e. without probabilistic inputs or outputs) aimed at explaining the behaviour of individuals in dynamic negotiations given any number of issues, in which the participants attempt to influence the outcome of a final and binding decision. Albeit different CDMM have produced acceptable predictions to actual final collective outcomes, both the data collection process and the interpretation of CDMM results require attention to the rather strict underlying assumptions in each of these models.

Our contribution is therefore twofold: (I) replication for systematic testing of the Challenge and Exchange CDMM assumptions, along with their requirements consisting of the Compromise, Mean and Median models, using an ABM framework; and (II) insights gained from these tests regarding the dynamics of  individual CDMM runs and their combinations using input from three empirical datasets.


Professor Cheryl Cott (University of Toronto, Canada)

Friday 26th October 2012

DL Computational Social Science Seminar Cott 2012

Using Social Network Analysis as part of an Edumetric Process in Interprofessional Teamwork

In this presentation I describe an applied social network project in which we used social network analysis as part of an edumetric process to develop interprofessional collaboration in primary health care teams in Ontario, Canada.  Edumetrics refers to the process of using research data as part of the educational process in order to stimulate a process of reflection amongst participants and to help them to develop a set of actions/goals to address issues raised by the data.  Members of 45 participating primary health care teams (e.g. family physicians, nurse practitioners, registered nurses, administrators and social workers) completed a questionnaire consisting of a measure of team functioning plus social network data on two types of ties with other members of the primary health care team (refer to and exchange information with).  Each team received a summary report containing: their team functioning scores; sociograms of the team for each type of tie plus a description of network density, centrality, strength of ties and reciprocity for each type of tie; and a set of reflection questions that the team could use as the basis for a reflective exercise in which all members of the team discuss the results and develop a set of actions/goals to address issues raised.  


Listings of Forthcoming Dynamics Lab Seminars are also posted at:

Geary Institute:

and CASL:


Previous Dynamics Lab presentations are also listed on the right-hand side of the website and below:

Loet Leydesdorff (University of Amsterdam, Netherlands)

“Meaning” as a sociological concept: A review of the modeling, mapping, and simulation of the communication of knowledge and meaning


Padraig Cunningham (CASL, UCD)

CLIQUE Research Programme


Frederic Amblard  (IRIT, University of Toulouse, France)

Agent-based models of the diffusion of (norms, opinions, information, attitudes …)

Abstract: From a computer scientist point of view, there are few general models that are used to render diffusion mechanisms in agent-based society. I will first present a global framework that enables to describe all these models. Then I will present what could be considered as building blocks for the construction of diffusion models in social simulation. I will then question this core library to see if there are some specificity of norms that would imply the use of one model rather than another


Jos Elkink (School of Politics and International Relations, UCD)


Nigel Gilbert (Surrey)

Agent Based Modelling and Simulation in the Social Sciences

Modelling complex social systems: opportunities and challenges

Abstract: Social systems are almost invariably complex and thus amenable to some of the techniques developed by physicists and mathematicians for the analysis of complex systems.  However, they are not the same as complex physical systems, and the differences are important.  In this talk, I shall compare the two, identify the features that make social systems special, and consider the implications of the differences.  This will lead me on to discuss some of the challenges that have to be faced in modelling social systems, especially using agent-based models, and to review the potential for this style of research.


Nigel Gilbert is Professor of Sociology and Director of the Centre for Research in Social Simulation at the University of Surrey, Guildford, United Kingdom. His research interests include innovation, agent-based modelling and research methods. He is the author of Simulation for the Social Scientist (with Klaus G. Troitzsch), (2005) Open University Press, and Agent Based Models (2007) Sage, and editor of the recently published four volume set, Computational Social Science (2010) Sage.



Petra Ahrweiler and other IRU colleagues at UCD

Innovation policy modelling from a complex systems perspective

Abstract: This presentation will outline a coherent framework for researching innovation in complex social systems. Complexity perspectives are highly relevant to innovation policy modelling. After a short introduction to innovation as a topic for research, the currently widely applied use of computational network analysis to investigate collaborative innovation arrangements will be discussed focusing on structural properties of these configurations. To include procedural and dynamic features of innovation in support of policy issues, a systemic perspective will be confirmed for social systems to connect agent-based modelling to the emerging integrated research program. Examples are given from current research projects on innovation policy modelling.