Body
Who can use this feature?
- Users with the "Knowledge Sets (Administrator)" permission (View and Edit) can create knowledge flows and use them to fill knowledge sets.
- An iPaaS license is required.
The system will execute the selected Knowledge Ingestion Flow, which retrieves a set of text records for indexing. This type of source can include almost any text content in a knowledge source.
This article covers the following topics:
This feature is available in two places, depending on your needs:
- Administration > Knowledge Flows
- Knowledge flows that are global to the system and used by both Conversational AI and Ticket Assist Agents.
Once you navigate here, a table of available knowledge flows will be displayed. From this view, you can create new flows, delete, or manage existing ones.
Similar to iPaaS applications, you can view your current flows, credentials, templates, and shared data on the Knowledge Ingestion Flows screen.
- On the Knowledge Ingestion Flows screen, click "Add." You can create an empty knowledge flow or import one from elsewhere.
- If you are creating an empty knowledge set, provide a name, with an optional short name and description.
- After clicking "Add New," the default knowledge flow will load.
- The default knowledge flow includes a table named KnowledgeSetRecords and an end step that returns the results from that table.
- To fill the table, set up a connection or file import and use that data.
- Note: The only required row is the Text row; adding more data can help the virtual service agent better respond to questions.
| Data Points |
Info |
| Text |
The body text of the article/resource that is being referenced. |
| ReferenceId |
A unique id which references this article/resource for updating existing content.
For TD KBs, this would be the Article ID.
|
| Reference |
The link that will be used by CAI as the hyperlink for the article. |
| ReferenceName |
The visible name or title of the resource/article. |
| Keywords |
You can filter knowledge sets by keywords in the knowledge query step.
If your articles have keywords or tags, you can add those to this list for easier filtering.
|
| Scope |
The scope is the greater repository where the article can be found.
For TD KBs, this would be something similar to eus_{environment}_{partitionID}_{BEID}
|
Adding a Knowledge Flow to a Knowledge Set
- In Conversational AI, go to Administration on the Top Navigation.
- Under the AI Knowledge Sets heading, select Knowledge Sets.
- Select the Knowledge set to which you want to add a web crawl source.
- In the Left Navigation Menu, select Sources.
- In the upper right, click the Add button to open a Create Knowledge Set Source dialog.
- In the dialog, provide a Name, a Description, and select the type of Knowledge Flow.
- Select the relevant knowledge flow you created.
- Click Add New.
In this example, we will create a Knowledge Flow that pulls in the Client Portal Service Catalog using a TeamDynamix connector.
- Add a connector and use the TDX Self Service Connector.
- Select your client portal app ID.
- Use the Get All Services action to retrieve the complete data catalog.
- Note: This action does not include all service data, so you will need to loop through these results.
- Add a loop that goes through the Get All Services > Response.
- In the loop, add a second TDX Self Service Connector.
- Use the Get Service action to retrieve complete data for a service.
- Select the same client portal app ID.
- In the Service ID, use the data picker to select the service IDs being looped through.
- Still in the loop, add the Add To data step to place the Get Service data in the KnowledgeSetRecords table.
- For the text column, use the long description.
- As mentioned, the other columns are optional but can improve accuracy.
- Once the flow is created, you can add it to a knowledge set using the steps outlined above.
