SearchUnify Release Notes (v23.10)

The v23.10 release for SearchUnify is available for from 26 October. The biggest change in this release is in ML Workbench, where you can not only look under the hood to learn how ML-powered features work, but also influence how Rich Snippets and Content Annotation work.

Control Panel in Rich Snippets (ML Workbench)

Rich Snippets has two tabs now instead of one: Playground and Control Panel. The latter is new.

In Control Panel you can configure which rich snippets are displayed on the search clients and the minimum score each rich snippet should have.

The delete function gives you the ability to remove undesired rich snippets. In Configure Threshold, you can supply the score. Only the rich snippets with scores that meet or exceed the threshold are displayed on the search results page. The threshold is changed with a slider. It means that you can tune the threshold, test, and save only when you're satisfied with the results.

Documentation: Machine Learning Workbench: Rich Snippets

Control Panel in Content Annotation (ML Workbench)

Control Panel in ML Workbench > Content Annotation allows you to view and edit content annotation requests submitted through Content Sources > Content Annotation.

You can choose how your content is annotated and view the strength of already-configured entity values found in a record. Two settings in the Control Panel—Threshold and Strongest Entity Rule—allow you to improve accuracy in content annotation.

Documentation: Machine Learning Workbench: Content Annotation

Quick Access to Machine Learning Models

A new tab, Models, in Machine Learning Workbench offers quick access to the Playground and Control Panel of features that are powered by machine learning. Currently, the access works for Rich Snippets and Content Annotation. Admins will see a Launch button for other features in the upcoming releases. Another big update in the future releases will be the Upload Your Own Model column where admins can test their machine learning frameworks in SearchUnify.

Documentation: Machine Learning (ML) Workbench

Recommendation Widget for Khoros

Back in the Mamba '23 release, a plug-and-play recommendations widget was introduced for Salesforce and website platforms. An advantage of the new widget was that it could be applied to any page, not just the search page. The same widget is now available on Khoros communities.

Documentation: Basic Search Client Configurations

Other Improvements