When you are adding an external knowledge base, there are a few points in the process where you need to provide a Jolt transformation spec that can be used to "transform" the response returned from the given request into the LivePerson KnowledgeAI article schema. This is necessary to map the external CMS' article data model to LivePerson KnowledgeAI.
If you were able to select your CMS provider from the Content Provider list, a default spec is available and provided at these points:
The default spec is pre-built for a default data structure. You can use the spec as is. Alternatively, if you've customized the CMS' data model (e.g., you've added a custom attribute), you can use it as a starter template.
If you weren't able to select your CMS provider from the Content Provider list, you'll need to write the spec from scratch, as no default spec is provided at these points:
Supported LivePerson attributes
When adding an external knowledge base without LivePerson AI, one step involves configuring the request for the on-demand content retrieval using your CMS’ query and answer API. This includes defining the transformation spec for the articles that are returned. The following LivePerson attributes are supported:
|summary||see above||no, but see note below|
|detail||see above||no, but see note below|
|contentURL||see above||no, but see note below|
|imageURL||see above||no, but see note below|
|audioURL||see above||no, but see note below|
|videoURL||see above||no, but see note below|
One of summary, detail, contentURL, imageURL, audioURL, or videoURL is mandatory.
Map content metadata
If you need to adjust or write the Jolt transformation spec, click Customize in the Add Knowledge Base window.
In the Transformation Spec editor (left panel), adjust or write the Jolt transformation spec using the examples and guidance.
In the Sample Data editor (middle panel):
- If the editor has default input, you can use it as is or replace it with your own, e.g., with the CMS' response payload.
- If the editor is empty, take your CMS' response payload and paste it in.
Click Validate Spec.
This verifies that the spec uses well-formed JSON and that the spec and sample input meet all LivePerson requirements. For example, if you've omitted a mandatory field, you'll see an error. If you've included a field that's not supported, you'll see a warning.
This step also populates the third, read-only Review Output panel. This panel illustrates the output when the sample input JSON is transformed using the spec. Behind the scenes, the output is JSON too. However, it's presented in a friendlier, record format, so it's faster and easier to evaluate whether the spec is working as you expect.
When you have finished with the spec, click Save.