I-Corps: CDnet: Community Detection in Large Networks

Major Goals: The goal is to explore the possibility of commercializing CDnet software product developed through the previous funded project. The product provides precise community detection in large network with superior performance. Through the funded projects, we studied the structure and dynamics of network communities and invented novel efficient methods for detection of network communities and building predictive models of behavior of groups of people. Although the actions of a particular individual may be too difficult to model, data mining and machine learning can be applied to large groups or ensembles, which can yield effective models and useful predictions. Our product CDnet has three main functions: (1) network community discovery and detection, 2) deep learning for network dynamics, and (3) a game-theoretic model of community evolution. This product has led to new insights into community structure of large networks, new theoretical models for understanding such structures and algorithms to harness them, and novel discoveries about connectivity structure of groups in networks. Such discoveries allow a company to explore its dynamics of network community structure and evolution. Accomplishments: We have conducted more than 100 interviews for potential customers. The project followed along the curriculum guidelines set by I-Corps, where the team will focus on the lessons taught and then quickly engages with industry/customer. The first task was to immediately identify customers/industries that utilize networks and would find advantages of having a CDnet. This approach will require determining the customers current limitations with existing technology, and whether switching the technology is justifiable. The second task was to determine the strategic market value of the technology. The goal is to determine that the proposed CDnet technology fit the customers needs and provide sufficient benefits to warrant the use of the CDnet.