Creating and Updating Models with Visual Modeling

Context Graph provides multiple methods of creating models. One way is to use the Write to Model stage or the Import to Model stage to create a model from entities and relationships that have been defined from incoming data.

Another way to create models is to perform visual modeling in the Relationship Analysis Client by manually adding entities and relationships from the Toolbox. You can build a model with two or more entities and one or more relationships. After that you could either use the same types of entities and relationships to populate your model, or you could create additional types of entities and relationships. For instance, if you were creating a model of insurance data that was comprised of patient information, you might need an entity type of "Patient" for patient names and an entity type of "Claim" for the illness or injury that instigated the claim. However, if your insurance model was comprised of patients and doctors, you would need to add a third entity type of "Doctor". The relationship type between patients and doctors might be "visited", and the relationship type between patient and claim might be "claimed" or "reported". Alternatively, you could have a relationship type between "Doctor" and "Claim" of "diagnosed". Using this data you could build a model that identifies potential fraud data if you see a doctor repeatedly diagnosing—or a patient repeatedly claiming—a particular injury or illness.

Within the visual modeling method in the Relationship Analysis Client, another way you can create and populate your model is by using metadata. First you add information types, in the form of entities. From there you can add relationships between those entities and completely populate the model from within the Relationship Analysis Client, manually adding properties for the entities and relationships, or you can instead populate it by mapping input fields to entities and defining relationships from within the Write to Model stage. Using the previous example, let's say you create entity types of "Patient", "Doctor", and "Claim" in the Relationship Analysis Client and then save your model with the name "Insurance Data". Then, you could go into Enterprise Designer and create a dataflow that has a Read from File stage going into a Write to Model stage. You could configure the Read from File stage to view an input file of insurance data that includes the following fields (among others): "TIN", "SSN", and "Dx". In the Write to Model stage, you would select "Insurance Data" in the "Model" field and then define entities by mapping input field "TIN" to type "Doctor", "SSN" to type "Patient", and "Dx" to type "Claim". After creating these entities, you would then define relationships between them. You could have "Patient_visited_Doctor" or "Doctor_treated_Claim" or "Patient_reported_Claim". After defining all the entities and relationships, you would run the job in Enterprise Designer, then go back to the Relationship Analysis Client, and you would see the "Insurance Data" model populated with the information that was in your input file, showing the relationships between the entities you just defined.

In addition to using the Relationship Analysis Client Toolbox to visually create models, you can use the Toolbox to add entities and relationships to an existing model. If you were looking at your "Insurance Data" model, you could create additional relationships between entities, or you could create new entities and add them to your model, linking them to other entities with existing or new relationships.

As you conduct visual modeling by creating a model or adding elements to your model, the Overview pane, the Model Details pane, the Model Components pane, and the Selection pane update accordingly, reflecting the properties in your model. This dynamic persistence enables you to use the tools of the Relationship Analysis Client in real time as you update your model.

For more information on visual modeling, see Creating Models and Editing Models.