Friday, December 29, 2017

Models from Teaching CSS

Most of the time when I teach a class instead of setting a final exam, I ask the students to carryout an end of semester research project. In my Introduction to Computational Social Science classes (both at the graduate and undergraduate level), this project entails the development of a computational model in an area of  interest to the student (or at the undergraduate level, students can opt to systematically explore someone else's model). The aim of this exercise is to cement what the students have (hopefully) learnt during the semester. I.e.:
  • to understand the motivation for the use of computational models in social science theory and research;
  • to learn about the variety of CSS research programs across the social science disciplines;
  • to understand the distinct contribution that CSS can make by providing specific insights about society, social phenomena at multiple scales, and the nature of social complexity.
Below you can see some of the outputs from these projects this last fall. The models range in type from agent-based models, cellular automata models to discrete event simulations (aka. queuing models) applied to a variety of topics from elephant poaching, artists and patrons, inheritance and wealth accumulation, the spread of religion, to that of looking at serving times at a Chipotle Mexican Grill.



 

Wednesday, December 13, 2017

Come Work with Us: 2 Postdocs in Urban Simulation

The George Mason University Department of Geography and Geoinformation Science within the College of Science, has an immediate opening for two postdoctoral fellows (up to 2-years), subject to budgetary approval. These positions will be part of the “Urban simulation” project team conducting research as part of the DARPA’s “Ground Truth” program, a network of DARPA-funded teams across the USA. The GMU team is directed by Andreas Züfle, Dieter Pfoser, and Andrew Crooks and supported by Carola Wenk at Tulane University. George Mason University has a strong institutional commitment to the achievement of excellence and diversity among its faculty and staff, and strongly encourages candidates to apply who will enrich Mason’s academic and culturally inclusive environment.

Postdoc 1

Responsibilities:
The primary job responsibilities of this position consist of the design, development and refinement of an agent-based simulation framework for urban areas. Using the existing Multiagent Simulation Toolkit (MASON) platform (written in Java), new agent logic will have to be implemented, thus creating agents that use socially plausible rules for mobility and interaction with other agents. A main goal is to create computationally efficient agent logic, thus allowing millions of agents to make decisions, find shortest paths between locations, and interact with their simulated world at the same time. For this purpose, implemented algorithms will need to be highly parallelizable, thus allowing to scale simulation via distribution among computing clusters located at GMU and Tulane. The successful candidate will also supervise graduate-level research assistants, collaborate with fellow scholars, and promote the department’s accomplishments through publications, presentations, and other public events.

Required Qualifications:
  • Ph.D. in computer science, data science, or closely related field;
  • Strong programming skills in Java;
  • Excellent written communication skills demonstrated by prior publications;
  • A track record that demonstrates the ability to work well with interdisciplinary research teams.

Preferred Qualifications:
  • Solid knowledge of graph algorithms;
  • Experience with Agent-Based Modeling and social science simulation;
  • Experience in design and implementation of software systems.

Postdoc 2

Responsibilities:
The primary job responsibilities of this position will be the design of an agent-based model based on the first principles underlying human needs, social interactions, and mobility to define socially plausible causalities. This model will contribute towards the design, development and refinement of an agent-based simulation framework for urban areas. Using the existing Multiagent Simulation Toolkit (MASON) platform (written in Java), new agent logic will have to be implemented, thus creating agents that use socially plausible rules for mobility and interaction with other agents. The successful candidate will also supervise graduate-level research assistants, collaborate with fellow scholars, and promote the department’s accomplishments through publications, presentations, and other public events.

Required Qualifications:
  • Ph.D. in computer science, data science, or closely related field; 
  • Experience with Agent-Based Modeling and social science simulation; 
  • Excellent written communication skills demonstrated by prior publications; 
  • A track record that demonstrates the ability to work well with interdisciplinary research teams.

Preferred Qualifications:
  • Strong programming skills in Java;
More Information: https://jobs.gmu.edu/postings/42109

Monday, December 11, 2017

Come work with us

The  Department of Computational and Data Sciences is currently looking for 2 Assistant Professors starting in the Fall of 2018.

Assistant Professor, Statistics and Visualization (Tenure-Track)

Required Qualifications:
Applicants must hold a Ph.D. in a closely related field from an accredited institution, and should have research interests congruent with the associated research centers listed below. Applicants should have a promising research record with a deep knowledge of and interest in computation as applied to statistical analysis and visualization, with a strong emphasis in one or more of the following fields: machine learning, Bayesian inference, and knowledge mining. Candidates should have a demonstrated ability or strong potential to attract funding and provide strong contributions to the continued growth of the academic programs.

More Information: https://jobs.gmu.edu/postings/41997


Assistant Professor, Data Science (Tenure-Track )

Required Qualifications:
Applicants must hold a Ph.D. in data science from an accredited institution or an aptly related field such as computer science, computational science, a physical science, etc., and should have research interests congruent with the associated research centers listed below. Candidates should have a demonstrated ability or strong potential to attract funding and provide strong contributions to the continued growth of the academic programs. Applicants should have a promising research record and a demonstrated deep knowledge in the expanding field of data science, with a strong emphasis in one or more of the following fields: data mining, knowledge mining, big data analytics, data engineering, and image analysis.

More Information: https://jobs.gmu.edu/postings/42005

About the Department: CDS is a rapidly growing department offering an undergraduate degree in Computational and Data Sciences and graduate programs in CSI (Computational Sciences and Informatics) and CSS (Computational Social Science). Interdisciplinary research directions are in modeling, simulation, data science, and computational social science. The successful candidate will benefit from a highly collaborative research environment that supports a wide array of shared facilities and dynamic research centers, such as the DataLab, the university-wide Center for Social Complexity, and the Department’s Center for Simulation and Modeling.


Friday, November 17, 2017

New Paper: Social Media and Cancer Campaigns

Continuing our work on geosocial analysis we recently had a paper entitled "Social Media Engagement with Cancer Awareness Campaigns Declined During the 2016 U.S. Presidential Election" published in  World Medical and Health Policy. In the paper we show through the analysis of Twitter and Google Trends, how public engagement with breast cancer and prostate cancer awareness months between 2015 and 2016 changed. Specifically we found that attention to breast cancer and prostate cancer declined in 2016 (during the U.S. presidential election), when compared to 2015. Based on our finding we suggest that future cancer education campaigns—and campaigns for other health issues and policies—would benefit from monitoring the broader issues producing social media engagement, and adjusting their timing or communication strategies to ensure that public engagement with their key messages remains strong even in a crowded social media marketplace. Below you can read the abstract to our paper, see some of our key findings along with the full citation to the paper.

Abstract:
Cancer awareness campaigns compete with other health and social issues for public attention. We examined whether public engagement with breast cancer and prostate cancer declined in 2016 during the U.S. presidential election compared to 2015 on Twitter and Google Trends. We found that attention to breast cancer and prostate cancer declined in 2016 before the election as compared to 2015 in Twitter posts and Google searches. The findings suggest that cancer information seeking behavior, passive exposure to health communication, and active participation in social media about cancer all decreased during the peak weeks of the 2016 election season. Future health promotion initiatives and information dissemination efforts will benefit from monitoring the major issues garnering social media attention and then adjusting their timing or communication strategies to ensure that public engagement with their key policy messages remains strong when emerging news stories capture public interest.

Keywords: Twitter; breast cancer; prostate cancer

Twitter Traffic about Breast Cancer (a) and Prostate Cancer (b) in the Period October 2015–November 2016, and Comparison of October and November Traffic in 2015 and 2016 for Breast Cancer (c) and Prostate Cancer (d). All 2015 Data are Shown in Red and All 2016 Data are Shown in Blue.


Full Reference:
Vraga, E. K., Radzikowski, J., Stefanidis, A., Croitoru, A., Crooks, A.T., Delamater, P., Pfoser, D. and Jacobsen, K. H. (2017). Social Media Engagement with Cancer Awareness Campaigns Declined During the 2016 U.S. Presidential Election. World Medical and Health Policy, 9(4): 456–465. (pdf)

Saturday, November 04, 2017

Call For Papers: Spatial ABMs: Current Practices & Future Trends

Special Issue theme: Spatial Agent-Based Models: Current Practices and Future Trends

Editors: Alison Heppenstall (Leeds) and Andrew Crooks (GMU)

Journal: GeoInformatica

Impact rating: 2.3

Overview

Over the last decade the agent-based modeling (ABM) paradigm has provided a new lens for understanding the effects of interactions of individuals and how through such interactions macro structures emerge, both in the social and physical environment of spatial systems. However, such a paradigm has been hindered due to computational power and a lack of large fine scale datasets. Within the last few years we have witnessed a massive increase in computational processing power and storage, combined with the onset of Big Data. Today geographers find themselves in a data rich era. We now have access to a variety of data sources (e.g., social media, mobile phone data, etc.) that tells us how, and when, individuals are using spaces. These data raise several questions: can we effectively use them to understand and model spatial systems as complex entities? How well have ABM approaches lent themselves to simulating the dynamics of spatial processes? What has been, or will be, the influence of Big Data on increasing our ability to understand and simulate spatial systems? What is the appropriate level of spatial analysis and time frame to model spatial phenomena? This special issue will concentrate on the best of current practice and future trends. We are interested in papers that will introduce the reader to:
  • Applications: Well-developed and transparent applications; 
  • Methodological innovations: use of ‘big data’; machine learning methods; calibration and validation methods within agent-based models; 
  • Thought pieces: What is the future of ABM? What do ABMs need to achieve to become as accepted similar to methods from climate change? How have and can agent-based models be used for policy making? 

Indicative deadlines 
  • Abstract (250 words): December 8, 2017 
  • Full Paper: April 30, 2018 
PLEAS E NOTE: 

Abstracts must first be submitted directly to the guest editors via email:

Further submission will then be invited based on the content assessed in the abstract. 

Full papers need to be between 5000 – 7000 words in length. Details of the journal submission requirements can be found at : http://www-users.cs.umn.edu/~shekhar/service/geoinformatica/guidelines2.doc

Tuesday, October 31, 2017

Happy Halloween....

As today is Halloween, I thought I would write a brief post on zombies and how they can be used to demonstrate disease models (even the Centers for Disease Control and Prevention (CDC) has a post about Zombies preparedness). There are several good examples of using zombie outbreaks as demonstrations for the utility of modeling (or just showing how modeling concepts can be applied to the spread of zombies). 

These range from exploring  the spatial and temporal dynamics of a zombie epidemic (e.g. Sander and Topaz, 2014). To that of the work of Alemi et al. (2015), who produced a "danger map" of what would happen if the continental United States  was overrun with zombies (an interactive version is available here and shown below). In their paper, they demonstrate how epidemiological processes akin to a Susceptible-Infected-Recovered (SIR) model (which we have wrote about before) can be used to model the spread of zombies. As its a zombie model, the SIR changes to a SZR model (Munz et al., 2009), where:
 "S represents the susceptible population, the uninfected humans, Z represents the infected state, zombies, and R represents our removed state, in  this  case  zombies  that  have  been  terminated  by  humans (canonically  by  destroying  their  brain  so  as  to  render  them inoperable)." (Alemi et al., 2015)
Zombie Town USA

Below you can see an attempt of modeling a zombie outbreak (only the SI parts) in one of the buildings on the George Mason Fairfax campus utilizing NetLogo (you can download the model code from here).



More complex individual based models have also been created like the one shown below by Horio and Arrowsmith (2015) which was used to showcase how zombies can be used to describe complex adaptive systems and agent-based modeling.

https://www.informs.org/ORMS-Today/Public-Articles/October-Volume-42-Number-5/The-Pedagogy-of-Zombies


If readers know of any over good Halloween (horror) like models, please let us know.

References:
Alemi, A.A., Bierbaum, M., Myers, C.R. and Sethna, J.P. (2015), 'You Can Run, You Can Hide: The Epidemiology and Statistical Mechanics of Zombies', Physical Review E, 92(5): 052801.
Horio, B. and Arrowsmith, N. (2015), 'The Pedagogy of Zombies', OR/MS Today, 42(5).
Munz, P., Hudea, I., Imad, J. and Smith, R.J. (2009), 'When Zombies Attack!: Mathematical Modelling of an Outbreak of Zombie Infection', in Tchuenche, J.M. and Chiyaka, C. (eds.), Infectious Disease Modelling Research Progress, Nova Science Publishers, Hauppauge, NY, pp. 133-150.
Sander, E. and Topaz, C.M. (2014), 'The Zombie Swarm: Epidemics In The Presence of Social Attraction And Repulsion', in Smith, R. (ed.) Mathematical Modelling of Zombies, University of Ottawa Press, Ottawa, Canada, pp. 265-300.

Friday, October 13, 2017

AAG2018: Innovations in Urban Analytics

Call for Papers, AAG2018: Innovations in Urban Analytics

We welcome paper submissions for our session at the Association of American Geographers Annual Meeting on 10-14 April, 2018, in New Orleans.

Session Description

New forms of data about people and cities, often termed ‘Big’, are fostering research that is disrupting many traditional fields. This is true in geography, and especially in those more technical branches of the discipline such as computational geography / geocomputation, spatial analytics and statistics, geographical data science, etc. These new forms of micro-level data have lead to new methodological approaches in order to better understand how urban systems behave. Increasingly, these approaches and data are being used to ask questions about how cities can be made more sustainable and efficient in the future.

This session will bring together the latest research in urban analytics. We are particularly interested in papers that engage with the following domains:
  • Agent-based modelling (ABM) and individual-based modelling;
  • Machine learning for urban analytics;
  • Innovations in consumer data analytics for understanding urban systems;
  • Real-time model calibration and data assimilation;
  • Spatio-temporal data analysis;
  • New data, case studies, demonstrators, and tools for the study of urban systems;
  • Complex systems analysis;
  • Geographic data mining and visualization;
  • Frequentist and Bayesian approaches to modelling cities.

Please e-mail the abstract and key words with your expression of intent to Nick Malleson (n.s.malleson@leeds.ac.uk) by 18 October, 2017 (one week before the AAG abstract deadline). Please make sure that your abstract conforms to the AAG guidelines in relation to title, word limit and key words and as specified at: http://annualmeeting.aag.org/submit_an_abstract. An abstract should be no more than 250 words that describe the presentation’s purpose, methods, and conclusions.

For those interested specifically in the interface between research and policy, they might consider submitting their paper to the session “Computation for Public Engagement in Complex Problems” (http://www.gisagents.org/2017/10/call-for-papers-computation-for-public.html).

Key Dates
  • 18 October, 2017: Abstract submission deadline. E-mail Nick Malleson by this date if you are interested in being in this session. Please submit an abstract and key words with your expression of intent.
  • 23 October, 2017: Session finalization and author notification.
  • 25 October, 2017: Final abstract submission to AAG, via the link above. All participants must register individually via this site. Upon registration you will be given a participant number (PIN). Send the PIN and a copy of your final abstract to Nick Malleson (n.s.malleson@leeds.ac.uk). Neither the organizers nor the AAG will edit the abstracts.
  • 8 November, 2017: AAG session organization deadline. Sessions submitted to AAG for approval.
  • 9-14 April, 2018: AAG Annual Meeting.

Session Organizers

Saturday, October 07, 2017

Generation of Realistic Mega-City Populations and Social Networks for ABM


At the upcoming 2017 Annual conference of the Computational Social Science Society of the Americas, Annetta Burger, Talha Oz, William Kennedy and myself have a paper entitled: "Generation of Realistic Mega-City Populations and Social Networks for Agent-Based Modeling". 

In the paper we discuss some of our current work of generating synthetic human populations with realistic social networks with respect to the New York mega-city and surrounding region. Below you can read the abstract of the paper and see our workflow along with some initial results. The full reference to the paper and a link to the pdf can be found at the bottom of the post.


Abstract:
Agent-based modeling is a means for researchers to conduct large-scale computer experiments on synthetic human populations and study their behaviors under different conditions. These models have been applied to questions regarding disease spread in epidemiology, terrorist and criminal activity in sociology, and traffic and commuting patterns in urban studies. However, developing realistic control populations remains a key challenge for the research and experimentation. Modelers must balance the need for representative, heterogeneous populations with the computational costs of developing large population sets. Increasingly these models also need to include the social network relationships within populations that influence social interactions and behavioral patterns. To address this we used a mixed method of iterative proportional fitting and network generation to build a synthesized subset population of the New York megacity and region. Our approach demonstrates how a robust population and social network relevant to specific human behavior can be synthesized for agent-based models. 

Keywords: Agent-based Models, Geographical Systems, Population Synthesis, Social Networks, Megacity.





Full Reference: 
Burger, A., Oz, T., Crooks, A.T. and Kennedy, W.G. (2017). Generation of Realistic Mega-City Populations and Social Networks for Agent-Based Modeling, The Computational Social Science Society of Americas Conference, Santa Fe, NM. (pdf)


Monday, October 02, 2017

Call for Papers – Computation for Public Engagement in Complex Problems

Call for Papers – Computation for Public Engagement in Complex Problems: From Big Data, to Modeling, to Action 



We welcome paper submissions for our session(s) at the Association of American Geographers Annual Meeting on 9-14 April, 2018, in New Orleans.  

Session Description: In line with one of the major themes of this conference, we explore the opportunities and challenges that geo-computational tools offer to support public engagement, deliberation and decision-making to address complex problems that link human, socioeconomic and biophysical systems at a variety of different spatial and temporal scales (e.g., climate change, resource depletion, and poverty). Modelers and data scientists have shown increasing interest in the intersection between science and policy, acknowledging that, for all the computational advances achieved to support policy and decision-making, these approaches remain frustratingly foreign to the public they are meant to serve. On one hand, there is a persistent gap in the public’s understanding of and reasoning about complex systems, resulting in unintended and undesirable consequences. On the other hand, there is significant public skepticism about the knowledge generated by the modeling community and its ability to inform policy and decision-making.

We invite theoretical, methodological, and empirical papers that explore advances in geo-computational approaches, including part or all the process to address complex problems: from data collection and analysis, to the development and use of models, to supporting action with data analysis and modeling. We are interested in any work that contributes towards the overall goal of supporting public engagement and action around complex problems, including—but not limited to—the following topics:
  • epistemological perspectives; 
  • extracting behavioral rules from novel and established data sets; 
  • innovative applications of complex systems techniques, and 
  • addressing the challenge of complex systems model calibration and validation. 

Please e-mail the abstract and key words with your expression of intent to Moira Zellner (mzellner@uic.edu) by October 18, 2017 (one week before the AAG abstract deadline). Please make sure that your abstract conforms to the AAG guidelines in relation to title, word limit and key words and as specified at: http://annualmeeting.aag.org/submit_an_abstract. An abstract should be no more than 250 words that describe the presentation’s purpose, methods, and conclusions.

 Timeline summary: 
  • October 18, 2017: Abstract submission deadline. E-mail Moira Zellner (mzellner@uic.edu) by this date if you are interested in being in this session. Please submit an abstract and key words with your expression of intent. 
  • October 23, 2017: Session finalization and author notification. 
  • October 25, 2017: Final abstract submission to AAG, via the link above. All participants must register individually via this site. Upon registration you will be given a participant number (PIN). Send the PIN and a copy of your final abstract to Moira Zellner. Neither the organizers nor the AAG will edit the abstracts. 
  • November 8, 2017: AAG session organization deadline. Sessions submitted to AAG for approval. 
  • April 9-14, 2018: AAG Annual Meeting.  

Organizers:

Monday, September 25, 2017

Talk: ABM For Simulating Spatial Systems: How Are We Doing?

Last week I gave a talk to Department of Computational and Data Sciences at George Mason University colloquium series entitled to "Agent-based Modeling For Simulating Spatial Systems: How Are We Doing?". This was an extension of a talk Alison Heppenstall and myself gave at the 2016 International Congress on Agent Computing. As we live stream and store these talks on YouTube, I thought I would share this one here (apologies about the sound quality). As normal, any thoughts or comments are most welcome.

Abstract:
While great advances in modeling have been made, one of the greatest challenges we face is that of understanding human behavior and how people perceive and behave in physical spaces. Can new sources of data (i.e. 'big data') be used to explore the connections between people and places? In this presentation, I will review the current state of the modeling geographical systems. I will highlight the challenges and opportunities through a series of examples that new data can be used to better understand and simulate how individuals behave within geographical systems.

Monday, September 18, 2017

Agent-Based Modeling Chapter

In the recently published "Comprehensive Geographic Information Systems" edited by Bo Huang, Alison Heppenstall, Nick Malleson and myself have a chapter entitled "Agent-based Modelling"1. Within the chapter, we provide a overview of agent-based modeling (ABM) especially for the geographical sciences. This includes a section on how ABM emerged i.e. "The Rise of the (Automated) Machines", along with a discussion on what constitutes an agent. This is followed with steps to building an agent-based model, including: 1) the preparation and design; 2) model implementation 3) and how one goes about evaluating a model (i.e. verification, calibration and validation and how these are particularity challenging with respect to spatial agent-based models). We then discuss how we can integrate space and GIS into agent-based models and review a number of open-source ABM toolkits (e.g. GAMA, MASON, NetLogo) before concluding with challenges and opportunities that we see ahead of us, such as adding more complex behaviors to agent-based models, and how "big data" offers new avenues for multiscale calibration and validation of agent-based models.  If you are still reading this, below you can read the abstract of the paper and find the full reference to the chapter.

Abstract:
Agent-based modeling (ABM) is a technique that allows us to explore how the interactions of heterogeneous individuals impact on the wider behavior of social/spatial systems. In this article, we introduce ABM and its utility for studying geographical systems. We discuss how agent-based models have evolved over the last 20 years and situate the discipline within the broader arena of geographical modeling. The main properties of ABM are introduced and we discuss how models are capable of capturing and incorporating human behavior. We then discuss the steps taken in building an agent-based model and the issues of verification and validation of such models. As the focus of the article is on ABM of geographical systems, we then discuss the need for integrating geographical information into models and techniques and toolkits that allow for such integration. Once the core concepts and techniques of creating agent-based models have been introduced, we then discuss a wide range of applications of agent-based models for exploring various aspects of geographical systems. We conclude the article by outlining challenges and opportunities of ABM in understanding geographical systems and human behavior.

Keywords: Agent-based modeling; Calibration; Complexity; Geographical information science; Modeling and simulation; Validation; Verification.





Full Reference
Crooks, A.T., Heppenstall, A. and Malleson, N. (2018), Agent-based Modelling, in Huang, B. (ed), Comprehensive Geographic Information Systems, Elsevier, Oxford, England. Volume 1, pp. 218-243 DOI: https://doi.org/10.1016/B978-0-12-409548-9.09704-9. (pdf)

1. [Readers of this blog might of expected the chapter would be about Agent-based Modeling, but its still worth a read!]

Thursday, August 17, 2017

Big Data, Agents and the City


In the recently published book "Big Data for Regional Science" edited by Laurie Schintler and  Zhenhua Chen, Nick Malleson, Sarah Wise, and Alison Heppenstall and myself have a chapter entitled: Big Data, Agents and the City. In the chapter we discuss how big data can be used with respect to building more powerful agent-based models. Specifically how data from say social media could be used to inform agents behaviors and their dynamics; along with helping with the calibration and validation of such models with a emphasis on urban systems. 

Below you can read the abstract of the chapter, see some of the figures we used to support our discussion, along with the full reference and a pdf proof of the chapter. As always any thoughts or comments are welcome.


Abstract:
Big Data (BD) offers researchers the scope to simulate population behavior through vastly more powerful Agent Based Models (ABMs), presenting exciting opportunities in the design and appraisal of policies and plans. Agent-based simulations capture system richness by representing micro-level agent choices and their dynamic interactions. They aid analysis of the processes which drive emergent population level phenomena, their change in the future, and their response to interventions. The potential of ABMs has led to a major increase in applications, yet models are limited in that the individual-level data required for robust, reliable calibration are often only available in aggregate form. New (‘big’) sources of data offer a wealth of information about the behavior (e.g. movements, actions, decisions) of individuals. By building ABMs with BD, it is possible to simulate society across many application areas, providing insight into the behavior, interactions, and wider social processes that drive urban systems. This chapter will discuss, in context of urban simulation, how BD can unlock the potential of ABMs, and how ABMs can leverage real value from BD.  In particular, we will focus on how BD can improve an agent’s abstract behavioral representation and suggest how combining these approaches can both reveal new insights into urban simulation, and also address some of the most pressing issues in agent-based modeling; particularly those of calibration and validation.

Keywords: Agent-based models, Big Data, Emergence, Cities.

The growth in Agent-based modeling -from search results of Web of Science and Google Scholar.

Hotspots of activity of Tweeter Users: Tweet locations and associated densities for a selection of prolific users.

Full Reference:
Crooks, A.T., Malleson, N., Wise, S. and Heppenstall, A. (2018), Big Data, Agents and the City, in Schintler, L.A. and Chen, Z. (eds.), Big Data for Urban and Regional Science, Routledge, New York, NY, pp. 204-213. (pdf)

Sunday, August 13, 2017

Predicting the Evolution of Narratives in Social Media


Building on our work on narratives and social media at the 15th International Symposium on Spatial and Temporal Databases (SSTD'17) we have a paper entitled: "Predicting the Evolution of Narratives in Social Media." In the paper we discuss briefly the challenges that social media poses with respect to understanding narratives and propose a framework that could be used to develop simulation models to predict the spread and evolution of narratives by blending the social, spatial and contextual dimensions of online narratives that are contextually informed by past events. Below you can read the abstract to our paper along with a link to the paper itself.
Abstract. The emergence of global networking capabilities (e.g. social media) has provided newfound mechanisms and avenues for information to be generated, disseminated, shaped, and consumed. The spread and evolution of online information represents a unique narrative ecosystem that is facilitated by cyberspace but operates at the nexus of three dimensions: the social network, the contextual, and the spatial. Current approaches to predict patterns of information spread across social media primarily focus on the social network dimension of the problem. The novel challenge formulated in this work is to blend the social, spatial, and contextual dimensions of online narratives in order to support high fidelity simulations that are contextually informed by past events, and support the multi-granular, reconfigural and dynamic prediction of the dissemination of a new narrative.



Full Reference:
Schmid, K. A. Zufle, A., Pfoser, D., Crooks, A.T., Croitoru, A. and Stefanidis, A. (2017), Predicting the Evolution of Narratives in Social Media, in Gertz, M., Renz, M., Zhou, X., Hoel, E., Ku, W.-S., Voisard, A., Zhang, C., Chen, H., Tang, L., Huang, Y., Lu, C.-T. and Ravada, S. (eds.) Advances in Spatial and Temporal Databases: Proceedings of the 2017 International Symposium on Spatial and Temporal Databases, Springer, New York, NY., pp. 388-392 (pdf)

Saturday, August 05, 2017

Spatial Agent-based Modeling to Explore Slum Formation Dynamics


In the newly published book edited by Jean-Claude Thill  and Suzana Drajicavic entitled:  "Geocomputational Analysis and Modeling of Regional Systems" Amit Patel, Naoru Koizumi and myslef have a chapter which explores some of our work with respect to modeling slums in India. The chapter is titled:  "Spatial Agent-based Modeling to Explore Slum Formation Dynamics in Ahmedabad, India." In which we report some of the work we did with pertaining to our sponsored NSF Project: "An Integrated Simulation Framework to Explore Spatio-temporal Dynamics of Slum Formation in Ahmedabad, India". Below you can see the abstract for the chapter along with some of the figures and a link to the project page.


 "More than 900 million people or one third of the world’s urban population lives in either slum or squatter settlements. Urbanization rates in developing countries are often so rapid that formal housing development cannot meet the demand. In the past decades, international, national and local development communities have taken several policy actions in an attempt to improve the living conditions of people within slums or to eradicate them completely. However, such policies have largely failed and slum-free cities have remained a distant goal for many developing countries. This chapter argues that for informed policymaking, it is important to investigate questions related to slum formation such as: (1) How do slums form and expand? (2) Where and when are they formed? (3) What types of structural changes and/or policy interventions could improve housing conditions for the urban poor? In order to address these questions, this chapter develops a geosimulation model that is capable of exploring the spatio-temporal dynamics of slum formation and simulating future formation and expansion of slums within cities of the developing world. Our geosimulation model integrates agent-based modeling (ABM) and Geographic Information System (GIS), methods that are often applied separately to explore slums. In our model, ABM simulates human behavior and GIS provides a spatial environment for the housing market. GIS is also used to analyze empirical data using spatial analyses techniques, which is in turn used to validate the model outputs. The core of this framework is a linked dynamic model operating at both micro and macro geographic and demographic scales. The model explores the collective effect of many interacting inhabitants of slums as well as non-slum actors (e.g. local government) and how their interactions within the spatial environment of the city generate the emergent structure of slums at the macro scale. We argue that when empirical data is absent, geosimulation provides useful insights to study implications of various policies. The goal of this framework is to develop a decision support tool that could allow urban planners and policymakers to experiment with new policy ideas ex-ante in a simulated environment. We calibrate and validate the model using data from Ahmedabad, the sixth largest city of India, where 41% of its population lives in slums. This is one of the first attempts to develop an integrated and multi-scalar analytical framework to tackle slum issues in the developing world at multiple spatial scales."
Keywords: Slums Agent-based modeling India Geosimulation

Integrated Simulation Framework
Slum Locations and Slum Sizes in Ahmedabad, 2001


Spatial Sprawl Experiment

Full Reference:
Patel, A., Crooks, A.T. and Koizumi, N. (2018). “Spatial Agent-based Modeling to Explore Slum Formation Dynamics in Ahmedabad, India” in Thill J.C. and Drajicavic, S. (eds.), Geocomputational Analysis and Modeling of Regional Systems, Springer, New York, NY, pp 121-141. (pdf)

Further details of the model and project can be found here. As normal any thoughts and comments are most welcome.

Monday, July 31, 2017

Travel Times, cost distances and more in NetLogo

Just a short post to highlight Rohan Fisher's excellent website demonstrating and sharing a number of NetLogo models. One such example is shown below, which integrates GIS cost distance analysis to explore access to services via travel times (click here to read the full paper).
 



In other examples, Rohan shows how NetLogo can be used to explore the spread of fire or how Cane toads can colonize new areas as shown in the movie below. More information about Rohan's work can be found at: https://rohanfisher.wordpress.com/ and his YouTube channel.





Friday, June 30, 2017

From Cyber Space to Physical Space Disease Outbreaks


At the upcoming 2017 International Conference on Social Computing, Behavioral-Cultural Modeling and Prediction and Behavior Representation in Modeling and Simulation  conference (or SBP-BRiMS 2017 for short), Xiaoyi Yuan and myself will present a paper entitled: "From Cyber Space Opinion Leaders and the Diffusion of Anti-vaccine Extremism to Physical Space Disease Outbreaks". In the paper we explore how online discussions with respect to vaccinations can potentially impact on the spread of a disease. Below you can read the abstract to our paper, see the basic model logic and movie of a single simulation. If you are interested in finding out more about the model or running it yourself, you can do so here: https://www.openabm.org/model/5509/


Measles is one of the leading causes of death among young children. In many developed countries with high measles, mumps, and rubella (MMR) vaccine coverage, measles outbreaks still happen each year. Previous research has demonstrated that what underlies the paradox of high vaccination coverage and measles outbreaks is the ineffectiveness of “herd immunity”, which has the false assumption that people are mixing randomly and there’s equal distribution of vaccinated population. In reality, the unvaccinated population is often clustered instead of not equally distributed. Meanwhile, the Internet has been one of the dominant information sources to gain vaccination knowledge and thus has also been the locus of the “anti-vaccine movement”. In this paper, we propose an agent-based model that explores sentiment diffusion and how this process creates anti-vaccination opinion clusters that leads to larger scale disease outbreaks. The model separates cyber space (where information diffuses) and physical space (where both information diffuses and diseases transmit). The results show that cyber space anti-vaccine opinion leaders have such an influence on anti-vaccine sentiments diffusion in the information network that even if the model starts with the majority of the population being pro-vaccine, the degree of disease outbreaks increases significantly. 

Keywords: Agent-based modeling Information networks Infectious disease transmission.



Full Reference:  
Yuan, X. and Crooks, A.T. (2017), From Cyber Space Opinion Leaders and the Spread of Anti-Vaccine Extremism to Physical Space Disease Outbreaks, in Lee, D., Lin, Y., Osgood, N. and Thomson, R. (eds.) Proceedings of the 2017 International Conference on Social Computing, Behavioral-Cultural Modeling and Prediction and Behavior Representation in Modeling and Simulation, Springer, New York, NY., pp. 114-119. (PDF).

Monday, June 05, 2017

Comparing four modeling approaches using a Susceptible-Infected-Recovered (SIR) epidemic model

Over the years several modeling styles have been developed but often it is unclear what are the differences between them. In this joint post, we, (Yang Zhou and myself) would like to compare and contrast four modeling approaches widely used in Computational Social Science, namely: System Dynamics (SD) models, Agent-based Models (ABM), Cellular Automata (CA) models, and Discrete Event Simulation (DES). For a review of their undying mechanisms and core components of each readers are referred to Gilbert and Troitzsch's (2005) "Simulation for the Social Scientist"

To compare and contrast the differences in how these models work and how their underlying mechanisms generate outputs, we needed a common problem to test them against with the same set of model parameters. While one could choose a more complex example, here we decided to chose one of the simplest models we know. Specifically, we chose to model the spread of a disease specifically using a Susceptible-Infected-Recovered (SIR) epidemic model. Our inspiration for this came from the SD model outlined in the great book “Introduction to Computational Science: Modeling and Simulation for the Sciences” by Shiflet and Shiflet (2014) which was implemented in NetLogo from the accompanying website. For the remaining models (i.e. the ABM, CA, and DES) we created models from scratch in NetLogo. Below we will introduce how we built each model, before showing the results from the four models with the same set of parameters, which allows us to compare the results of the models. The source code, further documentation for the four models can be found over at Yang Zhou's website and GitHub page.


The System Dynamics Model

In the system dynamics model from Shiflet and Shiflet (2014), one person is infected at start. Infected people can infect susceptible people. The population of infected will always increase by (number of infected * number of susceptible * InfectionRate * change in time dt). The infected people may recover. The amount of people that will recover in an iteration is always equal to (number of infected * RecoveryRate * change in time dt). Figure 1 illustrates the system dynamics process while Figure 2 shows the SIR process as a flowchart.

Figure 1. System Dynamics process (source: Shiflet and Shiflet, 2014)



Figure 2. System Dynamics flowchart


The Agent-based Model

As in the case for the SD model, at the beginning of the simulation, one agent is infected. Agents are randomly distributed on the landscape, and in the beginning of each iteration, they turn to a random direction and move forward by one cell. During each iteration, an infected agent may infect other agents on the same cell. This is different from how the SD model works, specifically the probability of getting infected. In the SD model, the infection rate is the infection rate on the entire population. In the ABM, the probability of becoming infected is equal to the infection rate divided by the probability of an agent to be in the same cell, multiplied by the change in time. Each infected agent has a probability to recover in each time period, which equals to the recovery rate times the change in time. The equations in the ABM are the following:

Where P(same cell) = probability to be on the same cell, equals 1 divided by total number of cells; dt = change in time. Figure 3 illustrates the agent decision process while Figure 4 shows the display of the ABM

Figure 3. Agent-based Modeling: agent decision process

Figure 4. Display of the ABM. Green = susceptible. Red = infected. Blue = recovered.

The Cellular Automata Model

At the beginning of the simulation, one cell is infected. During each iteration (dt), the infected cell can infect other cells in its Moore neighborhood (i.e. 8 surrounding cells). The landscape will be a n by n square, and n is equal to the square root of the number of people to be created at the beginning of the simulation. Wrapping is enabled both horizontally and vertically. Similar to the ABM, we would like to map the probability of becoming infected to the one in the SD model. In the CA model, the probability of becoming infected is equal to the infection rate divided by the probability to be in the Moore neighborhood, multiplied by the change in time. Each infected cell has a probability to recover in each time period, which is based on the recovery rate multiplied by the change in time. The equations here are:


Figure 5 shows the changing process of the cells while Figure 6 shows the display of the CA model.

Figure 5. Cellular Automata cell changing process

Figure 6. Display of the CA model. Green = susceptible. Red = infected. Blue = recovered.

The Discrete Event Simulation Model

In a Discrete Event Simulation model (aka. queuing model), there are three abstract types of objects: 1) servers, 2) customers, 3) queues, which is quite different from the CA and ABMs.

So to implement a SIR model as a DES Servers are the processes of becoming infected and recovering. The durations people stay with the servers represent the process of becoming infected and becoming recovered. Customers are susceptible people to be infected, and infected people are waiting to recover. We assume there are two queues in this model. As susceptible objects (i.e. individuals) are created, queues for infection are formed while people are waiting to be infected. On the other hand, as people get infected, they form a second queue waiting to recover. During each iteration (dt), each object in queue has a probability to get become infected. Each infected agent object has a probability to recover which is equal to RecoveryRate. After agents recover, they enter the sink of recovered people. The equations can be written as follow:


While the whole process is illustrated in Figure 7.
Figure 7. Discrete Event Simulation process.

Results from the Implementations


Now that the models have been briefly described. We turn to how using the same set of parameters lead to different results. The default parameters being used in each model are: number of susceptible people at setup = 2500, Infection Rate = 0.002, Recovery Rate = 0.5, change of time (dt) = 0.001, and the numbers of people in each status are recorded. Since the SD model has no randomness and will always give the same result, it is run only once. Each of the other three models were run for 10 times (feel free to run them more if you wish), and then we took the average of the ten results and show them in Figure 8. The stop condition is that no individual left to be infected.

Figure 8. Results for the different models. Clockwise from top left: SD model, ABM, DES and CA



In the four models, we observe the same pattern: the number of susceptible people decreases, the number of infected people increases first and then decrease again, and the number of recovered people increase over time. However, each model realization also shows a lot of differences in how such patterns play out.

First of all, the SD model has the smallest number of iterations before no one is infected. The number of iterations shown on the graph are the average of the ten runs, since the runs range from smaller to larger numbers (except for the SD model, which only has one run). The SD model only took 17451 iterations to stop, while the ABM took 19145 iterations (on average), the DES model took 18645 iterations (on average). The CA model took the longest time on average for no more individuals to be infected, it took 25680 iterations (on average).

The results of the SD, ABM and DES models while appearing to be very similar to each other. In the sense, that the number of infected people increase fast at first and reaches a peak number of over 1500 at more than 2000 iterations (2272 for SD, 2403 for ABM, 2538 for DES). On the other hand, in the CA model, the number of infected people increases much slower due to the diffusion mechanism of the CA model and never reaches an amount as high as in the former models.

An important characteristic of the SD model is that there is no randomness in the model, so no matter how many times you run this model, you will get the same result. In the other three models, getting infected or recover always depend on a probability function, so there is difference in every run.

Furthermore, people in the SD model and the DES model are homogeneous, and everyone has the same probability to becoming infected or recovering from an infection, although these rates change over time, they do not vary among the different people in the population. On the other hand, in the ABM and the CA model, people (represented by moving agents or static cells) are heterogenous in the sense that they have different locations. Only susceptible people around an infected individual can be infected. It is interesting that when people can move around, like in the ABM, the result is similar to the SD model, though the ABM takes a little more time to recover (19145 iterations in ABM vs. 17451 iterations in SD). When people are static and the number of people on the same space is limited (one cell in one space in this case), like in the CA model, the infection process becomes slower and it takes longer for everyone to recover.

To test how the models are sensitive to a specific parameter we now present what happens if we increase the infection rate in each model from 0.002 to 0.02 and show the results shown in Figure 9. As to be expected as the infection rate increased, the number of susceptible people decrease at a much faster rate. However, the SD, the ABM, and the DES models are still similar to each other, while the infection in the CA model is slower. The average number of iterations for these models are: 15807 (SD), 15252 (ABM), 16937 (CA), 16677 (DES). By increasing the infection rate the total number of iterations of each model has decreased, with the CA model still taking the longest time to converge. The peak of infected people in each model are on average: 2363 people at 255 iterations (SD), 2310 people at 363 iterations (ABM), 2035 people at 1019 iterations (CA), 2340 people at 286 iterations (DES). The CA model takes a longer time and reaches a lower peak.

Figure 9. Results for the different models with infection rate = 0.02. Clockwise from top left: SD model, ABM, DES and CA.

These models are only simple examples of how a SIR model can be implemented in different modeling techniques, but in reality, if we were to model disease propagation in more detail we would need to consider many other things such as people could be both moving through space (i.e. traveling to work) and static (i.e. staying at home), and the capacity of each cell is always limited to some amount.


References: 
Gilbert, N. and Troitzsch, K.G. (2005), Simulation for the Social Scientist (2nd Edition), Open University Press, Milton Keynes, UK.

Shiflet, A.B. and Shiflet, G.W. (2014), Introduction to Computational Science: Modeling and Simulation for the Sciences (2nd Edition), Princeton University Press, Princeton, NJ.
More information about the models and to download them please visit Yang Zhou's website.