Techniques for generating a clustered representation of a network based on node data

A computer-implemented method comprising:

• receiving, at the computer and from an input device, an indication of a specified variable for which each node of a network is associated with a value;
• receiving, at the computer and from the input device, an indication of one or more specified criteria to select a final clustered representation of the network, the one or more specified criteria comprising a maximum degree of loss of information concerning the specified variable associated with the final clustered representation;
• performing an iteration of derivation of the final clustered representation, wherein performing the iteration comprises:
• calculating a degree of loss of information concerning the specified variable arising from each possible combination of two linked nodes of multiple nodes of a current clustered representation of the network into a single node to generate a next clustered representation of the network, wherein each node of the clustered representation represents at least one node of the network;
• selecting one of the possible combinations of two linked nodes of the current clustered representation that is associated with a smallest degree of loss of information concerning the specified variable to combine into a single node;
• determining whether to cease performance of iterations based on whether the smallest degree of loss associated with the selected combination exceeds the maximum degree of loss of information;
• effecting the selected combination of two linked nodes of the current clustered representation to generate the next clustered representation in response to a determination that the smallest degree of loss does not exceed the maximum degree of loss;
• and
• selecting the current clustered representation as the final clustered representation in response to a determination that the smallest degree of loss exceeds the maximum degree of loss;
• generating a visualization of the network based on the final clustered representation;
• and
• transmitting the visualization to a display.

An apparatus comprising:

• a processor component;
• a communications component for execution by the processor component to receive, from an input device, an indication of a specified variable for which each node of a network is associated with a value, and to receive, from the input device, an indication of one or more specified criteria to select a final clustered representation of the network, the one or more specified criteria comprising a maximum degree of loss of information concerning the specified variable associated with the final clustered representation;
• and
• an iterative collapse component for execution by the processor component to perform an iteration of derivation of the final clustered representation, wherein performing the iteration comprises:
• calculating a degree of loss of information concerning the specified variable arising from each possible combination of two linked nodes of multiple nodes of a current clustered representation of the network into a single node to generate a next clustered representation of the network, wherein each node of the clustered representation represents at least one node of the network;
• selecting one of the possible combinations of two linked nodes of the current clustered representation that is associated with a smallest degree of loss of information concerning the specified variable to combine into a single node;
• determining whether to cease performance of iterations based on whether the smallest degree of loss associated with the selected combination exceeds the maximum degree of loss of information;
• effecting the selected combination of two linked nodes of the current clustered representation to generate the next clustered representation in response to a determination that the smallest degree of loss does not exceed the maximum degree of loss;
• and
• selecting the current clustered representation as the final clustered representation in response to a determination that the smallest degree of loss exceeds the maximum degree of loss;
• and
• a rendering component for execution by the processor component to generate a visualization of the network based on the final clustered representation and to transmit the visualization to a display.