A cognition lets you design the logic needed to classify your historic, transactional data into a standardized format. When designing a cognition, you will use a design canvas to map the fields from your historic, input data to the fields of your selected, output classification family.
The mappings between input and output fields can be direct so that the value in the input field is written directly to the output field. Alternatively, the mappings can contain classification logic that is determined by the arrangement and use of the various Logic nodes. Classifier and Script nodes also allow you to use Python scripts in cognition mappings to leverage machine learning algorithms in the classification process.
The following image shows an example of a direct mapping within a cognition.
The Work Order Priority input field node is directly mapped to the Priority output field node. When the cognition is executed, values in the Work Order Priority field for the various work history events will be written directly into the Priority field for the corresponding, classified versions of the work history events.
The following image shows an example of a mapping with classification logic within a cognition.
The mapping is made of nodes and connections that define the classification logic for this output field. Specifically, the nodes in the mapping represent:
The input fields from your input data that contain historic, transactional data.
In this example, the Event Short Description, Event Long Description, and Order Priority Description nodes are the input fields.
The Logic nodes that add specific logic to classify the input data into a standardized format before it is written to the output field.
In this example, the Is A Failure? Classifier node is using the isAFailure.py script to produce a classification prediction of a standard value that is based upon the data provided in the input fields.
The output field in the selected classification family that stores your classified data.
In this example, the Failure? node is the output field.
When the cognition is executed, the data in the input fields for the various work history events is processed by the Is A Failure? Classifier node, and the classification prediction is stored in the Failure? field for the corresponding, classified versions of the work history events.
Tip: For additional examples, refer to the topics in the About Logic Nodes section of this documentation.
IMPORTANT: When you create a cognition, you should ensure that no other cognition is configured to modify the same set of data. This is important to avoid a scenario where multiple cognitions make conflicting changes to the same set of records.
Copyright © 2018 General Electric Company. All rights reserved.