Algorithms

The Algorithms function allows you to apply centrality to a model.

Centrality is a way of measuring the importance and significance of individual entities and relationships. When you run centrality algorithms, the higher the value, the more important the element.
  1. Click the Algorithm drop-down to select the kind of centrality measure you want to apply to your model:
    OptionDescription
    Betweenness Used to identify entities that control the information flow between different parts of the network.
    Closeness Used to identify entities that may have best access to other parts of the network and visibility of activities within the rest of the network.
    Degree Used to identify entities that have the most direct links to others.
    Influence Used to identify entities that have strong influence in the network due to their direct links to other highly active or well-connected entities.
  2. Select the direction in which you want to apply the algorithm:
    OptionDescription
    Incoming The results will be based on relationships coming into the entity.
    Outgoing The results will be based on relationships going out of the entity.
    Both The results will be based on incoming and outgoing relationships.
  3. If you are using an Influence algorithm, slide the Precision scale to determine how precise the results should be.
    A lower precision will return more accurate results, but the algorithm will run more slowly.
  4. If you are using a Closeness algorithm, click the appropriate button for the way in which you want results to be returned:
    OptionDescription
    Standard Results are based on the number of attachments, or relationships, an entity has as well as the reverse of the sum of shortest paths to each entity.
    Dangalchev Results are based not only on the number of entities linked to another entity but also the number of relationships in each of the linked entities.
    Opsahl Results are based on the sum of reversed shortest paths to each entity.
  5. Click the Use relationship property as weight if you want to measure how unfavorable a relationship is, and select the relationship property you want to use from the Property drop-down.
    In this case, a higher value indicates a negative association.
  6. Click the Low values are more significant box if you are using a relationship property as weight and that property is one where a lower value is considered better than a higher value.
    For example, if the property is some sort of ranking system, typically 1, or 1st, is the best value. Another example is if the property is distance, and you are trying to determine the shortest route: 5 miles is considered better than 10 miles.
  7. Click the Override default output property name check box if you want the output property name to be something other than the algorithm you selected. Then enter the new name in the Property field.
  8. Click Run.
  9. Click the Jobs tab to view job details after running the algorithm.

Completing this procedure provides the job ID, the model name, the algorithm used, the status of the job, the start time, and the end time. Be aware that the bigger the model, the longer it takes the algorithm to run; watch the Status column to determine if a job is still running.

Note: Results are not stored over time. If you close Relationship Analysis and reopen it, the information on the Jobs tab is cleared.