Link Analysis Terminology

The following table provides a hierarchically ordered list of terms that describe the concepts of Link Analysis.

Table 7-1 Link Analysis Glossary

Term Definition
node Business entity of interest (for example, institution, bank, tax ID,employee, customer, account, address, or household) involved in a link.
link Specific occurrence of an association between two nodes (for example, a wire transfer from one customer to another customer,or to the address of a beneficiary). Zero (0), one (1), or more links can exist between two nodes.
relationship Association between two nodes (for example, the two nodes are related in one or more transactions). Only a zero (0) or one (1)relationship can exist between two nodes.
network Collection of nodes that links connect.
primary node Node in the network that has the most links to other distinct nodes.
network definition Type of relationship that defines a network, the characteristics that describe a network, and the logic to derive those characteristics.
sourcelink dataset Dataset that represents a set of logic used to find links and nodes and the relationship between them.
Network Characteristics dataset Dataset that represents the characteristics or description of a network—the results of the definable characteristics. Only one Network Characteristics dataset can exist for each Network Definition.
input characteristic dataset Dataset that specifies the logic to derive the characteristics stored in the Network Characteristics dataset. You can use more than one input characteristic dataset in a Network Definition.
standard characteristics Properties of a network; for example, the number of nodes or average link weight. Within the Link Analysis process, fourteen standard characteristics are automatically captured and stored in the KDD_NTWRK table for a network.
definable characteristics Properties or descriptors of a network that the Data Miner derives. They are based either on the network itself or on the business data associated with the network. The number of definable characteristics for a network is unlimited as long as a table in the database contains the results.