The research at SCNARC is concentrated into three main areas: SCNARC only research, research in Trust in Distributed Decision Making Cross Cutting Research Initiative (Trust CCRI) and in Evolving Dynamic Integrated/Composite Networks Cross Cutting Research Initiative (EDIN CCRI). The mission of Collaborative Technology Alliance is to foster collaboration between different research areas and groups. SCNARC only research concentrates on issues specific to social and cognitive networks. Research in CCRIs concentrate on problems that cut across different network types: social and cognitive networks, information and communication networks. The research effort is organized in the following three projects. Note that the tasks in these projects involve only researchers from SCNARC.

S1: Value of Network Interactions

This project focuses on understanding and utilizing the value of network interactions. It addresses the challenges of analyzing and processing large-scale network in real time for finding the value and roles of information and people. Efficient graph algorithms are in high demand for social interaction mining, split-second decision making, context-aware personalized information filtering, summarization, routing, etc. We focus on fundamental issues in parallel analytic algorithms, high-performance computing, network indexing and compression, and fast retrieval of large networks, building a critical algorithmic foundation for all NS CTA tasks that require network data analysis. We also investigate market-based approaches to participatory sensing/monitoring/routing, in which people carrying (on themselves or their vehicles) wireless devices equipped with sensing capabilities (e.g. cell phones with cameras, sensors for pollution, chemical traces, etc.) agree to sense or monitor the environment and/or route messages. Finally, we study social relations between nodes, including relationships between friends, co-workers, family members, etc., to develop metrics for direct and indirect social relations. The metrics will be used in forwarding decision in opportunistic routing. The project is strongly linked to tasks in all three centers and CCRI Trust. Especially strong collaboration has been established with tasks in CNARC as well as INARC and IRC.

Task S1.1: Scientific Challenges of Large-Scale and Real-Time Network Mining Infrastructure (Lead: C. Y. Lin (IBM), Tong (IBM), Z. Wen (IBM), T. Brown (CUNY), N. Chawla (ND), J. Hendler (RPI), S. Aral (NYU))

Task S1.2: Market-Based Approaches to Social Network Based Participatory Monitoring & Routing (B. Szymanski (RPI), C.Y. Lin (RPI), A. Pentland (MIT))

Task S1.3: Socially-Driven Models of Human Mobility and Its Use in Delay Tolerant Routing (D. Lazer (NEU), C.Y. Lin (IBM), B. Szymanski (RPI))

S2: Adversary Social Networks: Detection, Evolution, and Dissolution

The overall goal of this project is to study adversary networks through the communication and information signature that such networks create during internal interactions. The broad research questions that we address in this project include identification of communities in a dynamic social network, especially hidden and informal groups, based on measurable interactions between the members. We also study relations between communities in terms of membership, trust and opposition or support and observe evolution and the stable cores of communities, especially anomalous and adversary communities or groups, and their relationships. Finally, we develop efficient strategies to dissolve communities in social networks, corresponding to adversarial communities with hostile, extremist and/or militant ideologies. This project has linked tasks both in Trust and EDIN as well as collaborative tasks in IRC and INARC.

Task S2.1: Adversary Social Networks: Detection, Evolution, Stability and Hierarchy (M. Magdon-Ismail (RPI), M. Goldberg (RPI), W.A. Wallace (RPI), N. Chawla (ND))

Task S2.2: Forming, Dissolving, and Influencing Communities in Social Networks (G. Korniss (RPI), B. Szymanski (RPI), C. Lim (RPI), Makse (CUNY), Z. Toroczkai (ND), A. Barabasi (NEU))

S3: The Cognitive Social Science of Net-Centric Interactions

A fundamental element of any human-centric network is the brain and the constraints that this “hardware” puts on human information processing and decision-making. Current models of human cognition are best at predicting normative cognition under conditions of complete information, low stress, and low time pressure. For the soldier in the field, such conditions are rarely met. Accordingly, this year’s task emphasizes time-stressed decision-making under conditions of partial and uncertain information with multiple sources (nodes) of information. Typically, the decision maker must gather information to decide on a course of action before his/her opponent can act. New information becomes available over time but delaying information in favor of gathering more complete information increases the likelihood that the opponent acts first. This year we introduce two paradigms that work over two different time and network scales. The paradigms will allow us to examine how variations in information sequencing, number of sources (nodes in the information network) of information, and types of sources (human or instrumented) influence human trust and decision-making performance in a networked environment. We will use cognitive models to predict behavior of an ideal human performer, measure human performance against these ideal models, and determine how feedback and training can be used to improve human decision-making behavior. There is a linked task in CCRI Trust to which we will contribute results and collaboration with other Trust CCRI and INARC tasks.

See the CogWorks Lab for more information.

Trust CCRI:

Trust in distributed decision making concentrates on a unified view of trust, one that integrates trust in all aspects of networks, social trust, trust in information and communication systems. The Trust CCRI is organized into two main projects that incorporate tasks that go across all the centers. The main aim of the CCRI is to foster collaboration between centers and different institutions to study the underlying science of trust. The projects in Trust are:

T1: Models of Propagation, Corroboration, and Provenance for Enhancing Trust: This project examines the fundamentals of modeling trust, metrics that quantify trust, and properties and dynamics that impact the metrics. We will develop a mathematical model of composite networks and a supporting framework that will serve as a unifying construct for specification of trust metrics. We will also develop models for provenance, corroboration, and argumentation that enable us to build systems that construct, convey, and deliver trustworthy OIC to the warfighter. Finally, we will develop models of propagation of trust that accommodate the dynamics of the network and the context in which trust decisions need to be made.

T2: Network Dynamics and Trust: This project focuses on the dynamics of trust and credibility in composite networks. In the first task, we will develop algorithms and protocols that support partial revocation of trust in the presence of fuzzy measures of distrust and support mutual revocation schemes with arbitration via a third party. Second, we will develop cognitive models of trust that support judgments of credibility and trust in information flows to support development and testing of novel user interface designs. We will also develop novel schemes for in-network storage that use the semantics of trust to influence data replication strategies that are resilient to the dynamics of trust relationship formation/dissolution and context. Finally, we will develop network-based indicators of trust in composite social and information networks through the mining of the social context of topics and sentiments in socially-diverse networks.

Trust SCNARC Participation:

Task T1.1: Models and Metrics of Trust in Composite Networks (Lead: W. A. Wallace, (RPI, SCNARC), S. Adali (RPI, SCNARC), M. Singh, (NCSU, CNARC), A. Singh (UCSB, INARC))

Task T1.2: Understanding the Dynamics of Corroboration (Lead: R. Govindan (USC, CNARC), M. Singh (NCSU, CNARC), B. Uzzi (Northwestern, SCNARC))

Task T1.4: An Argumentation-based Approach to Trust and Liability Estimation in Composite Networks (Lead: S. Parsons (CUNY, SCNARC), K. Haigh (BBN, IRC), K. Levitt (UCD, CNARC), M. Singh (NCSU, CNARC))

Task T1.5: Trust Propagation in Composite Networks (Lead: K. Haigh (BBN, IRC), Kawadia (BBN, IRC), Z. Wen (IBM, SCNARC), C.Y. Lin (IBM, SCNARC), F. Wu (UCD, CNARC))

Task T2.2: Cognitive Models of Trust (Lead: P. Pirolli (PARC, INARC), W. Gray (RPI, SCNARC), M. Schoelles (RPI, SCNARC), B. Suh (PARC, INARC), T. Hollerer (UCSB, INARC))

T2.4 Network Behavior-based Indicators of Trust in Composite Social and Information Networks (Lead: S. Adali (RPI, SCNARC), M. Magdon-Ismail (RPI, SCNARC), M. Goldberg (RPI, SCNARC), W. A. Wallace (RPI, SCNARC), J. Golbeck (UMD, SCNARC), P. Pirolli (PARC, INARC), B. Suh (PARC, INARC))


EDIN focuses on evolution and dynamics of integrated networks. Processes is such networks are governed by the structure of the network itself and its components, such as social and cognitive networks, information networks and underlying computer networks, as well as by the internal dynamics of those processes (or behaviors) governed by the activities of the entities operating in those networks. However, those entities are also capable to change the structure of the network. Hence, the integrated research in EDIN will take into account the strong feedback loop between the component networks, network structure and goals and needs of the entities that operate processes in the network. Gaining a fundamental understanding of such feedback loops under different kinds of dynamic processes operating int he network is instrumental for obtaining significant insights into the behavior of all kind of integrated networks, ranging form complex military tactical networks to social network in highly hierarchical organizational structure (e.g., industrial entity, military unit) or in a loosely connected, often purposefully hidden, adversarial networks. The projects in EDIN are:

E1: Modeling of Composite Networks: This project is organized into four tasks. The first aims to formally model composite/interacting networks to facilitate formal reasoning on such spatio-temporal structures. The second task focuses on developing a theory of composite graphs by developing mathematical techniques for gluing together multiple graphs derived from component networks and understanding their properties. The third task proposes to develop advanced mathematical techniques beyond graph theory to model group behaviors in social, information, and communication networks. The final task focuses on developing efficient techniques for discovering the structure of adversarial networks.

E2: Dynamics and Co-evolution of Composite Networks: This project is organized into three tasks. The first is focused on methods to model time-varying networks. The second focuses on one of the causes of network dynamics—mobility—and its impact on the topology of communications and information networks. The final task addresses the impetus of dynamics from a social networks perspective, and provides linkages between the different network types.

EDIN SCNARC Participation:

Task E1.1: Formal Modeling of Dynamic Networks and Network Interactions (Lead: K. Haigh (BBN, IRC), J. Hendler (RPI, SCNARC), J. Bao (RPI, SCNARC), A. Singh (UCSB, INARC), Dean (BBN, IRC))

Task E1.4: Adversarial Network Discovery by Generalized Random Walks (Lead: Z. Toroczkai (ND, SCNARC), D’Souza (UC Davis, CNARC), A. Vespignani (IU, SCNARC), N. Kawadia (BBN, IRC), G. Korniss (RPI, SCNARC), B. Szymanksi (RPI, SCNARC))

Task E2.3 Co-evolution of Composite Networks with Dynamic Processes and Communities (Lead: L. Adamic (UMich, INARC), L. Barabasi (NEU, SCNARC), N. Chawla (ND, SCNARC), J.J. Garcia-Luna-Aceves (UCSC, CNARC), J. Han (UIUC, INARC), O. Lizardo (UND, SCNARC), Z. Toroczkai (UND, SCNARC), A. Vespignani (IU, SCNARC))