It can be very useful to know about activities among individuals; for example, which individuals are associated with other individuals? When two or more individuals get together, is there an intended purpose? Who are the leaders or important individuals of a group? What is the organizational structure of the group? It can prove useful to have the capability to actually model the above types of interactions and associations. To an extent, this type of social research has been addressed by employing the disciplines of data mining and community generation.
A software suite was developed for identifying and analyzing social groups and communities from complicated data sources where explicit relationships are not yet known. Traditional data mining has focused on known relationships between people and transactions. Such methodologies, however, fail when trying to identify key associates in situations such as advanced money laundering schemes and terrorist organizations.
The Uni-Parity Data Community Generation (UDCG) paradigm and Link Discovery are based on Correlation Analysis, a new methodology to discover social groups. The purpose is to greatly reduce the time required for analysts to discover and analyze communities of interest and key figures where explicit relationships are not obvious.
A suite of mathematical models and software extracts pertinent data from large datasets (paper, electronic, and online), analyzes complex datasets, determines key groups within those datasets, and provides visualization tools. Key developments include a name resolution system for data extraction, and a mathematical approach that is insensitive to transaction errors. The system has successfully been validated using actual data from two large financial frauds: a $45 million money laundering and Ponzi scheme based in the US, and the Enron scandal.