In preparation for enabling Triage, one of the first steps is to train a supervised deep learning model. This involves feeding data to a machine learning algorithm to facilitate its learning process and enable it to make reliable predictions. The quality of the data used for training is very important. It is essential to ensure that the model is trained on good data, as training it with low-quality data could negatively affect its performance.
As a starting point, Triage models depend on data that includes specific fields and meets certain requirements, as outlined in the example below. Although there may be minor differences in naming conventions based on the ticketing system utilized, the fundamental requirements remain consistent across systems.
While exceptional cases may require additional fields, the following set of fields is generally adequate for the majority of models.
Desired Output: A .csv with the following columns -
- Data Column, e.g.
- Example, e.g.
|Ticket ID||° INC123456|
|Unique identifier of each individual ticket.||Yes|
|Ticket type||° Incident|
° IT Help
|The classification or categorization of a ticket||Yes|
|Created at||° 12/7/22 15:59||Ticket creation time||Yes|
|Predicted field||° assignment_group|
|The field Triage will be predicting||Yes|
|Display name||° Help Desk - 1|
° SAP Support
|The displayed name of the predicted field. This concept of display name vs internal name only applies to ServcieNow customers.||Yes|
|Internal Name (of Display Name)||° er2d29f29249fh93498h||A system-generated alphanumeric value that uniquely identifies a particular record, such as an incident, task, or any other entity within the ServiceNow platform.||If applicable|
|Short Description||° I need help with my VPN||Ticket Short Description||Yes|
|Description||° I need help. My VPN doesn’t seem to be working.||Ticket Description||Yes|
|Contact type / Route||° Bot|
|The channel or method through which a ticket is received or routed within the system. Triage accepts only digital channels (email, self service, bot, etc).||Yes|
|Indicates the state of progress or resolution of the ticket. Triage models are exclusively trained using tickets that are no longer open or require resolution.||Yes|
|Reassignment Count||° 0|
|The number of times a ticket has been reassigned to a different assignee or group.||If applicable|
|Initial group||° Help Desk - 1|
° SAP Support
|First assignee group or team that is initially responsible for handling a ticket.||If applicable|
|Location||° San Francisco|
|Geographical location associated with a ticket||If applicable|
Updated 3 months ago