Trails: Assessing Flows as Dynamic Networks
Abstract
Trails: Assessing Flows as Dynamic Networks Forecasting requires understanding the order in which events are likely to unfold. Currently, it is difficult to tell how people will move between various life options or services, or how ideas will move between groups, cities or countries. At the physical level, the physical geography and the built environment constrain the movement of people and objects. This enables route planning. However at the organizational, social and cultural level, there are fewer physical constraints. Rather movement is impacted by norms, preferences, and social networks. A general problem in trying to predict the movement of people or ideas among locations is the nature of the data. Trying to forecast movement of people and ideas, requires inferring what they will do from what they have done. This is a classic case of the Inverse problem faced by climatologists and weather forecasters. The inverse problem occurs when the direction of inference (infer today from yesterday) is in the opposite direction of causality (yesterday causes today). A feature of inverse problems is that the paths of change are not pre?defined – hence there is a discovery component. A second feature is that there are more variables than there is data; so methods need to use non?data?hungry techniques like the Lhasso procedure. This proposal is to develop a new methodological approach for predicting the movement of people or ideas among locations, using a novel dynamic network approach. The proposed technique essentially merges social networks and Markov modeling techniques. Lhasso estimation techniques are used to reduce the space of variables when identifying which elements co?move together in topic assessment. The anticipated result is a new predictive modeling capability for dynamic network data that can be characterized as trails of movement. Key challenges to be addressed include: link identification and weighting, metrics, and automation. The new modeling capability will be tested using four sets of data to test for generalizability and scalability. Military Relevance: Forecasting likely change supports mission planning at both the tactical and strategic level. This work lays the ground work for a novel approach to forecasting movement at the population and sub?group level. In particular this work supports predicting the movement of groups between organizational services, and the diffusion of ideas. The former is relevant for a) improving veterans care, and b) assessing usage patterns for provided care such as usage of disaster services. The latter is relevant for MISO, and Intelligence operations. If the same technology could support both, there would be a reduction in training requirements. Scientific Impact: This work creates a new dynamic network methodology that will enable new research to be conducted on social change. This work also supports the development of “big data” metrics for social network research that go beyond the current volumetrics approach to indepth understanding of the underlying structure. PI: Kathleen M. Carley, Carnegie Mellon University Requested Total Funds: $114,836. 1 year effort
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 08, 2016
- Source ID
- N000141512563
Entities
People
- Kathleen Carley
Organizations
- Massachusetts Institute of Technology
- Office of Naval Research
- United States Navy