OpenAI can now be configured as a Text Embedding Provider for Semantic Search.

OpenAI offers two powerful third-generation embedding models promising lower costs, higher multilingual performance, and stronger overall performance compared to the predecessor models.

If your organization is already utilizing OpenAI's ChatGPT or other OpenAI Platform APIs, now you can take advantage of OpenAI's Vector Embeddings capability to deliver semantically more relevant search results through Liferay's semantic search feature (BETA) without the need to sign up for another service.

 

The Data Set Manager sorting tab has changed its purpose. From now on, the sorting options created under this section will be added to an Order Drop Down that is added to the Management Bar. This way, users will be able to sort by different criteria, even if those criteria are not visible in the Data Set items.

Admins can now leverage Data Set actions to send targeted requests to a specified endpoint. This enables dynamic updates to persisted objects, such as changing a user's status to "Inactive." While the request body content is currently static, this powerful feature provides a flexible way to manage and modify data within your system.

Data Sets now offer even greater customization power. Admins can directly add manual query parameters to the endpoint URL, enabling fine-grained control over data retrieval.

This flexibility allows you to:

• Filter Data: Apply specific filters to narrow down the results
• Select Fields: Choose the exact fields you want to retrieve
• Flatten Responses: Simplify complex data structures
• Search for Specific Terms: Find relevant information quickly

By tailoring the endpoint response, admins can provide a more curated and user-friendly experience for end-users.

Behind the Scenes:

The platform seamlessly reconciles these manual parameters with those generated by:

• Admin Configuration: Fields and Actions tab settings
• End-User Interactions: Filter and sort options selected in the UI

This ensures a consistent and efficient data retrieval process.

On top of that, admin users can also add or map fields manually to be displayed in the Data Set, so they will be able to bring the information back for those endpoints that don’t have the fields properly typed in the response.