#NetworkScience
In this Newsblast Dr. Chris Arney shares his perspective on network science and its role in information systems. He discusses the shift from physical to information science and how network science gives us the tools necessary to study how to collect, manage, analyze, interpret, and present relataional data. Click here to read the complete article.
You will also find links to some of the most viewed posts from the Central Node over the last month.
#NetworkScience
It has not been a surprise that the continued research into Somali pirate networks has revealed greater complexity that originally anticipated; however, what has been surprising is the degree of that complexity. While there had always been the assumed ties of pirate crew and gangs, of clans and subclans, and perhaps even of regional affiliation, it has been the interplay of individuals, identities, and roles that has contributed most to the complex picture that Somali piracy offers. Recent data gathered by my cadet assistants Eric Warren, Jonathan Ambriz, and Ian McWilliams has offered an exceptional example of the multiple relationships involved when tracing the piracy off of Somalia.
Abdi Saed Bafe Looyan
Of the current agents traced, Abdi Saed Bafe Looyan stands as the perfect example of the complex ties within the network. Looyan is a pirate, but has gained most of his notoriety as a negotiator for other pirate gangs. He has been connected to the capture and ransom of over a dozen ships, including the MV Samho Dream, MV Frigia, and MV Victoria. These actions in turn allow us to tie him to pirate financiers such as Abdi Yare and Mohamed Abdi (a.k.a. Garaad) through his work negotiating ransoms for ships taken by their crews. However, his links to Garaad do not end there; he and Garaad are cousins and share membership in the Majerteen subclan of the Hawiye Clan. Through this link, we can also begin joining Mohamed Abdi and Looyan to the larger network of Mejerteen pirates. This is also enhanced through geographic factors, with Looyan and Garaad operating within distinct areas and these can then be linked to other pirates that work within those regions, specifically Garowe and Galkayo. As such, even through a single individual, albeit a prominent one, the piracy network has managed to become one of intermixed family relationships, clan linkages, economic ties, and proximity to form a multimodal network that can be analyzed to offer a revealing view of how these networks form and function.
#NetworkScience
PI: COL John Graham and LTC Ian McCulloh (U.S. Military Academy)
In 2007, the Network Science Center began a research project sponsored by The United States Army Research Institute to examine new management tools for the information age. In particular, to what degree can email and blackberry communication data provide insight into informal networks within a military chain of command? In addition, what type of longitudinal behavior can be expected of military group dynamics? Will they continually evolve, or will they converge to a stable structure.
What was accomplished:
We obtained IRB approval to issue blackberries to every officer assigned to the Eisenhower Leadership Development Program (ELDP), and cadets serving in regimental chain of command. This provided a closed pool of leaders who had reason for daily email and phone interaction in the course of their duties. We had access to the formal organizational structure and could conduct surveys to understand the informal network structure of the participants.
Social network email data were collected over the course of the year-long program. The cadets in the chain of command were surveyed weekly during the Spring semester, reporting the time they spent with other group members, their ranking of friendship, their trust, and for select members, their geo-spatial data. The multiplex relations among the agents in the network were analyzed against each other and longitudinally.
Multi-agent simulation was used to model agents within an Infantry company and generate longitudinal social networks. Changes were introduced at known points in time in the simulation, in order to create a controlled environment to explore network change detection. Multiple change detection approaches were compared.
Methods for handling periodicity in data were developed using wavelets and the Fourier transform. In addition, change detection methods were automated in a social network package, ORA, under a grant from ARL.
Why it’s important:
Email provides insight into informal group dynamics. Email can be easily monitored to provide commanders information on the informal networks within their units. These networks can be monitored in real time. Network change detection algorithms have been developed and demonstrated to be effective in monitoring organizational behavior for significant change.
Methods for handling periodicity provide more accurate real-time monitoring of networks over time, with fewer false alarms.
Prior to organizations successfully carrying out some activity or event, there exists a change in the social network of the organization as members plan and resource the activity. Social network change detection may allow an analyst to detect the network change prior to the event, allowing a commander to get inside the decision cycle and influence the outcome.
This approach can also be used with adversarial network data to enhance intelligence of terrorist organizations.
#NetworkScience
Last week I had the privilege of joining the Somali Pirate Network team led by Dr. Charles Thomas from the history department here at West Point on a very exciting project! Team members include Cadets Eric Warren and Jonathan Ambriz, who have collected data from 2012 pirate activity off the coast of Somalia. For example, they have collected information on who was involved in various ship robberies and the locations of these events. They also determined clan and sub-clan relationships between pirates, as well as family ties. The pirates themselves assume one or more roles: for example some are financiers, others are hostage negotiators.
My role in this project will be to create networks from this data the team has so carefully collected. I will be linking agents by relationships and by involvement in the same pirate event. We hope to use this data to understand the pirate social structure and determine key players. I am looking forward to working on this important project and will post updates as our work progresses.
#NetworkScience
The genesis of the forthcoming project on Warlord networks has its roots in Dr. Luke Gerdes and my mutual interest in Violent Non-State Actors (VNSAs). While this category includes organized crime groups, secessionist fronts, and mercenaries, few have been so synonymous with the United States’ efforts in the developing world as those grouped under the tag of the ‘Warlord Insurgencies.’ However, during our discussions of these phenomena, we quickly came to the conclusion that the existing literature on them was spotty and vague at best and self-contradictory at worst. Despite the ubiquity of ‘Big Men’ labeled as Warlords, there were few studies that really dug past a surface reading of the armed group and its overt power structures. We both agreed on the ability of Network Science to not only penetrate far deeper than previous studies, but also trace the nuances involved in these groups, from shifting political alliances to informal hierarchies of power by constructing an agent-level network of multiple warlord insurgencies.
With this in mind, the next step was identifying possible case studies. As can be guessed, the normal difficulty in dealing with VNSAs is actually gathering agent-level data on them. These groups tend to be armed, unpredictable, and not especially friendly to scrutiny. As such, the best possible case studies were those that had ostensibly ended, were recent enough to still be relevant to policy-making circles, but at least somewhat accessible to researchers. Given those parameters, the recently concluded warlord insurgences of the First Liberian Civil War (1989-1996) and the Sierra Leone Civil War (1991-2002) could serve as excellent foundational studies to build a larger portfolio upon. Both conflicts featured one or more strong central figures characterized as Warlords (most notably Charles Taylor in Liberia and Foday Sankoh in Sierra Leone); expansive networks of formal and informal combatants; transnational lines of support and supply; and had occurred in the recent past. However, the key in terms of choosing these conflicts was the presence of the Truth and Reconciliation Commission (TRC) at the cessation of the wars. The presence of the TRCs meant that former combatants could speak freely of their experiences, as they had gone through the process of being formally reconciled to the state. In addition, the records and transcripts of the TRC proceedings can be used to supplement the information we gather firsthand from informants.
The project is now in its final planning stages before an initial research trip to Liberia this summer. Dr. Gerdes, two cadets, and I will be traveling to Monrovia in order to conduct semi-structures interviews, gather publicly disseminated information, and expand our dataset of former combatants. As the study develops, more details will be provided via the Central Node.
#NetworkScience
Last week, I traveled to Hamburg, Germany, to participate in the 33rd annual Sunbelt Conference of the International Network for Social Network Analysis (INSNA). I presented two papers. The first paper, which I co-authored with MAJ John Kendall (an instructor in West Point’s Department of Social Sciences), demonstrates that ethnic groups in Afghanistan maintain political cooperation, despite a long history of ethnic conflict. However, the paper also suggests that ethnic groups try to manipulate the political system for their own ends, because after his election to the Afghan presidency in 2005, Hamid Karzai was able to extend the extent of direct political influence held by his ethnic group, the Pashtuns.
The second paper, which I authored alone, examines the role that data transformation plays in determining which individuals play important structural roles in clandestine networks. My research shows that analysts need to think very carefully when selecting an appropriate method to transform two-mode data (e.g. agent x event interactions) to one-mode data (e.g. agent x agent interactions) – a step that is necessary in order to implement standard measures of network centrality – because different transformation methods often produce very different results. In effect, researchers can incorrectly determine which individuals are most important in a two-mode network simply by selecting the wrong data transformation method. If analysts pick the wrong method, they may pick the wrong people. My paper also offered a new data transformation method, the Median Additive Projection Process (MAPP), which I designed to specially suit the needs of analysts working with agent x event data on clandestine networks, such as those comprised of criminals or terrorists. I hope that MAPP will help analysts with the difficult task of identifying key functional leaders in violent networks.
#NetworkScience
There are numerous applications which require the ability to take certain actions (e.g. distribute money, medicines, people etc.) over a geographic region. A disaster relief organization must allocate people and supplies to parts of a region after a disaster. A public health organization must allocate limited vaccine to people across a region. In both cases, the organization is trying to optimize something (e.g. minimize expected number of people with a disease).
In April, at the IEEE Network Science Workshop, we introduced “geospatial optimization problems” (GOPs) where an organization has limited resources and budget to take actions in a geographic area. The actions result in one or more properties changing for one or more locations. There are also certain constraints on the combinations of actions that can be taken.
In the paper, we study two types of GOPs – goal-based and benefit-maximizing (GBGOP and BMGOP respectively). A GBGOP ensures that certain properties must be true at specified locations after the actions are taken while a BMGOP optimizes a linear benefit function. We show both problems to be NP-hard and prove limits on approximation for both problems. We present integer programs for both GOPs that provide exact solutions. We also correctly reduce the number of variables in for the GBGOP integer constraints. For BMGOP, we present the BMGOP-Compute algorithm that runs in PTIME and provides a reasonable approximation guarantee in most cases.
A pre-print of the paper can be found online at http://arxiv.org/abs/1303.4632.
#NetworkScience
Part 7: Micro-Finance & A Tech Meet-Up
Mr. Dan Evans and Dr. Charles Thomas visited Kampala, Uganda in support of an ongoing Network Science Center project developing new models of entrepreneurial networks. As additional data sets are collected, and new models are developed, we will compare and contrast the models based on the development of new network classification methods. After the networks have been classified, we will develop a method to identify driver nodes enabling policy makers to influence the existing network in order to cause it to evolve towards a more desirable state.
During the course of our trip we had several chance encounters that added greatly to our knowledge as well as to our data collection efforts.
The first of these was a last minute meeting with Fred Iga Luganda, the Director of the Makerere University Business School Micro-Finance Center. The Center is a research organization aimed at fostering financial inclusion and fighting poverty through research, training, consultancy, Internship and an information hub for policy makers and researchers. The MFC is developing innovative solutions based on the results of their research in order to design better products and delivery mechanisms to serve the unbanked.
Fred was a gracious host; he explained his Center’s goals and current work and we had the opportunity to discuss our ongoing research project with him. Additionally, we discussed potential ways to collaborate in the future.
Kampala Tech Meet-up at Hive CoLab
The second was an invitation to attend a Kampala tech meet-up at Hive CoLab entitled, “The New Face of Entrepreneurship” co-sponsored by Global Business Labs, an international organization that initially set up a business incubator model at the Stockholm School of Economics and has recently established similar labs in Namibia, Botswana and Uganda.
Kampala Tech Meet-up at Hive CoLab
#NetworkScience
Dr. Cyril Cassagnes, an EECS visiting professor from the University of Bordeaux Computer Science Research Laboratory in France, spoke to researchers from the West Point Network Science Center on his work with tactical mobile clouds during a recent brown bag luncheon. He noted that soldiers at the lowest tactical level continue to have an increasing need to share information with a growing and diverse set of partners. Humanitarian assistance and disaster relief (HADR) events such as the Tōhoku earthquake and tsunami in Japan illustrate the need for international partnership. The proposed tactical mobile cloud implies multiple layers of sometimes conflicting goals and constraints:
1. Upper layer: Provides services such as sensor data collection, computation or information-sharing, all within a cloud architecture.
2. Middle layer: The primary responsibility of the middle layer is to maintain connectivity of the swarm and the secondary mission is to present an abstraction of the swarm to the cloud. This layer has to deal with the swarm motion to maintain a unique virtual network. Additionally, it will act as a management layer to arbitrate between events in the swarm and the requirements that support cloud services.
3. Lower layer: This includes the set of robots and motion strategies needed to accomplish a goal or task, such as coverage and connectivity maintenance for the cloud users. Additionally, environmental events may invoke other motion strategies through sensing and avoiding as well as capability constraints (e.g. energy consumption or availability of sensors).
Leveraging and extending previous research in formal models of distributed systems, Dr. Cassagnes will use finite-state automata and the related formal language to model the proposed system. Additionally, the formal system description will be used to implement a simulation. The goal then is to verify the consistency between the formal analysis and the simulation. The vision is that this formal work will enable researchers to build mobile clouds with predictable performance that can be optimized.
For more information on Dr. Cassagnes’ research, please see “Tactical Mobile Clouds.”
#NetworkScience
Part 6: The Survey and More Data Collection
Mr. Dan Evans and Dr. Charles Thomas visited Kampala, Uganda in support of an ongoing Network Science Center project developing new models of entrepreneurial networks. As additional data sets are collected, and new models are developed, we will compare and contrast the models based on the development of new network classification methods. After the networks have been classified, we will develop a method to identify driver nodes enabling policy makers to influence the existing network in order to cause it to evolve towards a more desirable state.
During our visit to Kampala, we were greatly assisted by the staff of Hive CoLab in arranging to meet with entrepreneurs throughout the city. We were fortunate to be able to interview many of the entrepreneurs who are members at Hive CoLab and Brian Ndyaguma, Hive’s Project Manager, also introduced us to mangers at some of the other business incubators located in Kampala. Based on these introductions, we were able to interview additional local entrepreneurs.
Mara LaunchPad
Our third set of interviews outside of Hive CoLab occurred at Mara LaunchPad located in Ham Towers across the street from the main entrance to Makerere University. At Mara LaunchPad we met Delia Dean, the LaunchPad General Manager. Delia is also a business development consultant.
Mara LaunchPad was founded and funded in 2010 by the Mara Foundation which was established by the founder and Director of Mara Group, Ashish J. Thakkar. Ashish is a Ugandan-born entrepreneur who grew a small computer trading operation into a diversified conglomerate with approximately $100 million in revenues. Mara is now involved in such diverse businesses as real estate and tourism, financial services, information and communications technology, renewable energy. and manufacturing. The group’s operations span 16 countries on four continents, including Asia.
Ashish is driven by his concern about youth unemployment and is interested developing mentorship and guidance for young entrepreneurs. LaunchPad links each entrepreneur with a mentor and focuses on accounting and financial training.
Mara Launchpad was founded in partnership with Angels Finance Cooperation (AFC), a business support organization started by three Ugandan school friends who we were fortunate to meet. Being young Ugandan entrepreneurs themselves, the members of the AFC team are able to directly relate to the needs and challenges of the members of the Launchpad.
Mara LaunchPad has an open-plan layout with modern furniture. Members are able to take advantage of the fast Wi-Fi internet connection, lounge area, and conference room. The businesses under incubation at Mara LaunchPad are more diverse than those at the other incubators that we visited. Their start-ups include manufacturing companies, call center operations, and agriculture, in addition to tech start-ups. Mara has 2 time periods per year where entrepreneurs can apply to join. Once accepted, the entrepreneur signs a contract. LaunchPad typically seeds approximately $2,000- $4,000 per company and takes an equity stake with a three to five year time frame. Additionally, each firm pays rent ranging from $35-$125 per month depending on the size of their space. The incubation goal is 24 months. Mara can house up to 40 businesses at one time and their Innovation Center has room for 50-60 individuals.
In addition to the shared office space, Mara Launchpad runs a speaker program and hosts many external events. The speaker program brings keynote professionals to talk to entrepreneurs on technical issues of relevance to them, such as taxation and venture capital. The Young Entrepreneur Club meetings of the Entrepreneur Launchpad mentorship program also take place at the Mara Launchpad.
Next: Micro-Finance & A Tech Meet-Up
