Summer Research Projects

Over the summer, Arie Croitoru and myself took part for the first time in the George Mason University Aspiring Scientists Summer Internship Program (ASSIP). We worked with three very talented high-school students who over the course of the seven and a half week program produced some excellent research around the areas of agent-based modeling (ABM), volunteered geographical analysis (VGI), social media and text analysis. An overview of their work can be seen below in the posters and abstracts that the students produced at the end of the internship.

End of Summer Research Poster Display

In the first project, Kevin Geng and Varun Talwar explored how online news stories propagate around the world in a project entitled: "MediaPulse: A System Prototype for News Media Aggregation and Analytics". We would also like to thank Trevor Thrall for his assistance and guidance in this project. Below you can read the abstract and see the poster from this work.

"The news media play a major role in shaping the opinions and beliefs of people around the globe. Alongside more traditional media distribution modes, such as the printed press, in recent years the Internet has begun to play a more important role in news distribution. With the emergence of online news, information can now be disseminated across the globe in real-time. By monitoring such online sources, we can, for the first time, obtain valuable insight into the propagation of news around the world over time, and understand how such media is both produced and consumed. However, obtaining and analyzing this data on a massive scale has proven to be challenging. To address this challenge, we present a novel system for the collection and analysis of news articles. Our system has the capability to extract both metadata and text from a large array of online news sources, and analyze it with respect to themes, locations, time, and language. In order to showcase the utility of our system, we selected 440 prominent news sources across the globe, and monitored their RSS feeds every hour using our system. Through this process we collected roughly 30,000 articles per day over the duration of the study period. To demonstrate the analytical capabilities of our system, we present a case study analysis of coverage of the Islamic State of Iraq and Syria (ISIS) using the system that we developed. In particular, we analyze ISIS-related news using both their content and metadata to show how news propagates over time and space, and explore how the sentiment of the coverage varies."

In a second project Rohan Suri developed an agent-based model to explore the spread and  containment of Ebola in a project entitled "A novel computational agent based model for the spread and containment of Ebola Virus Disease". Below is the abstract, poster and a example model run from this project. More information about the model can be found here.

"During the Summer of 2014, the countries of Sierra Leone, Guinea, and Liberia were devastated by an Ebola Virus Disease (EVD) epidemic.  Although it killed more than 10,000 people, little is known about EVD dynamics in a macro population. While various attempts have been made to better understand EVD dynamics, such past attempts at modeling EVD exclude an explicit spatial scale, implied general mixing, and did not consider human-to-human interactions. In view of these limitations, this research aims to develop a novel computational agent based model (ABM) to investigate spatial and temporal EVD spread, and to study the effectiveness of control and prevention measures for EVD. In this model, OpenStreetMap (OSM) data was used to construct the physical environment (e.g., road networks), and a realistic population for the three countries was generated from Landscan data and previous surveys. EVD spread was modeled through explicit agent-to-agent interaction and the use of a Suspected-Infected-Exposed-Recovered (SEIR) model"

It was a great learning experience from our side by participating in the ASSIP Program.

Slides from Lipari: GIS and Agent-Based

I have just come back from the Lipari School on Computational Social Science which was focusing on Algorithms, Data, and Models for Social and Urban Systems. While there I gave a class entitled "GIS and Agent-based modeling" split over two parts. Below you can see the abstract for my class and the slides.

Abstract: Understanding and modeling human behavior is not as simple as it sounds. This is because humans do not just make random decisions, but base actions upon their knowledge and abilities. Moreover, one might think that human behavior is rational, but this is not always the case: decisions can also be based on emotions (e.g. happiness, anger, fear). Emotions can influence decision-making by altering our perceptions about the environment and future evaluations. The question therefore is how we model human behavior? Over the last decade, one of the dominant ways of modeling human behavior in its many shapes and forms has been through agent-based modeling (ABM). In this workshop I will provide a general overview of what agents are, why there is a need for agent-based models for studying cities, and how it links to how we believe societies operate through ideas of complexity theory. I will sketch out how geographical information systems (GIS) can be used to create spatially explicit agent-based models, before reviewing a range of applications where agent-based models have been developed using geographical data. The workshop concludes with an overview of challenges modelers face when using agent-based models to study geographical problems with a special emphasis on cities, and identify future avenues of research relating to big data and social network analysis.

   

In addition to giving my own class I was also fortunate to attend a diverse set of classes by several other people including  Alessandro Mei, Carlo Ratti, Claudio Cioffi Revilla, Paolo Ferragina, Wulf Daseking and Daniele Quercia. All of which explored how urban systems can be studied from a variety of disciplines. One of the highlights of the summer school for me was interacting with the students and learning about their own research. The school was excellent and the location was perfect for learning and relaxing. 

Computational Social Science for Urban Dynamics

I just came back from a workshop organized by the JASON's that was focusing on decision making from complex modeling. During the workshop I gave a talked entitled "Computational Social Science for Urban Dynamics"

In the talk I introduced, computational social science  (CSS) and why one should use it for studying urban systems. Then I went over a collection of agent-based modeling (ABM) applications from basic examples to more complex ones such as pedestrian movement, border security. I then discussed the rise of crowdsourcing and how this has made new kinds of geographic data available to the computational CSS community. New forms of data range in their characteristics and purpose. One example is Volunteered Geographical Information (VGI), were users purposely contribute Geographic Information (GI) as in the case of OpenStreetMap; another is Ambient Geographic Information (AGI), where the intention of contributors is not necessarily to provide GI, but GI can be derived, as from Twitter.  I then  discuss the opportunities that crowdsourced data provides for ABMs, specifically focusing on how such information gives us a new lens to study the micro interactions of individuals. Through as series of examples I  demonstrated how such data can be integrated into geographically explicit ABMs. By building on these examples I  showcased how the spatial environment and agent populations can be built using crowdsourced information, and highlight how agent behaviors can be informed and validated by such information. Lastly I discussed the challenges associated with this program of research: using such data is not without its difficulties, including gathering or accessing the data, storing the data, analyzing the collected data, and assessing its validity. Overall this talk provides a brief overview of the current state of crowdsourced data-informed ABM.

Please note many of the models and papers cited in the presentation can be found under research or publications sections of this website. Also I would like to acknowledge and say thank you to the GeoSocial team at Mason along with current and former students for providing material and much food for thought for this presentation.

A Semester with Spatial Agent-based Models

With the spring semester over, I thought I would show some of the final agent-based modeling projects that were carried out in CSS 645: Spatial Agent-based Models of Human-Environment Interactions. As always I was quite impressed the models and we had a plethora of topics ranging from mobile agent-based models, shopping, pick pocketing, route finding, travel to work, the spatial spread of information, deer management, urban growth etc... What is interesting is that while the majority of models are implemented in NetLogo, more and more are being done in Python.

Something new for this semester is we also tried to reproduce a published model. Below you can see 3 examples of such work. Click here to see a previous post on reproduction and replication. 

Leveraging Crowdsourced data for Agent-based modeling: Opportunities, Examples and Challenges

This week I am attending the AAG Annual Meeting in Chicago. While here, we organized 3 sessions entitled "Geosimulation and Big Data: A Marriage made in Heaven or Hell?" in which I presented a paper, co-authored with Sarah Wise: "Leveraging Crowdsourced data for Agent-based modeling: Opportunities, Examples and Challenges." The abstract is below:

The rise of crowdsourcing has made new kinds of data available to the  geographical community. New forms of data range in their characteristics and purpose. One example is Volunteered Geographical Information (VGI), were users purposely contribute Geographic Information (GI) as in the case of OpenStreetMap; another is Ambient Geographic Information (AGI), where the intention of contributors is not necessarily to provide GI, but GI can be derived, as from Twitter. While much progress has been made in utilizing these new sources of data in GIScience, they have only recently begun to be integrated into agent-based models (ABM). This paper will discuss the opportunities that crowdsourced data provides for ABMs, specifically focusing on how such information gives us a new lens to study the micro-interactions of individuals. Through as series of examples we will demonstrate how such data can be integrated into geographically explicit ABMs. By building on these examples we will showcase how the spatial environment and agent populations can be built using crowdsourced information, and highlight how agent behaviors can be informed and validated by such information. We further discuss the challenges associated with this program of research: using such data is not without its difficulties, including gathering or accessing the data, storing the data, analyzing the collected data, and assessing its validity. Together, this work provides a brief overview of the current state of crowdsourced data-informed ABM. 

If you like what is written above, you can have a flick through the slides from the talk or check out one of the movies: