Synonym Sets now also supports the following languages (locales) out of the box:

• Portuguese (Brazil): pt_BR

• Portuguese (Portugal): pt_PT

• Italian: it_IT

• Arabic (Saudi Arabia): ar_SA

• Catalan (Spain): ca_ES

• Catalan (Andorra): ca_AD

• Catalan (Spain, Valencian): ca_ES_VALENCIA

• Chinese (China): zh_CN

• Chinese (Taiwan): zh_TW

• Dutch (Netherlands): nl_NL

• Dutch (Belgium): nl_BE

• Finnish (Finland): fi_FI

• French (France): fr_FR

• French (Canadian): fr_CA

• German (Germany): de_DE

• Hungarian (Hungary): hu_HU

• Japanese (Japan): ja_JP

• Swedish (Sweden): sv_SE

Administrators are advised to review the Elasticsearch connector configuration and remove any customized field mappings and settings added previously for the above locales from the Override Type Mappings and the Additional Index Settings as now they are provided by the platform out-of-the-box. (Requires a Full or Concurrent Reindex!)

Improved Disk Space Utilization for Elasticsearch and OpenSearch Deployments

To reduce disk space usage on the search engine side, we have optimized default field mappings which no longer use stored fields. Instead, the platform uses the fields parameter as the default document field retrieval method.

Important Considerations for Administrators:

• Review Advanced Settings: Administrators should carefully examine the "Additional Type Mappings" and "Override Type Mappings" settings within the search engine connector configuration.

• Compare with Defaults: After reviewing the advanced settings, compare the existing mappings to the new defaults to identify potential adjustments. While stored fields are still supported, they are not recommended.

Potential Impact on Custom OSGi Modules:

While most users should experience seamless transitions, developers of custom OSGi modules that utilize Liferay's Java Search APIs may encounter changes in search response behavior. This could occur if the modules rely on specific fields being included in the search response. In such cases, developers may need to make adjustments to their modules to ensure correct functionality.

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.

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.

We added the permissions assigned to Custom and Modifiable Object Entries to the Objects Entries APIs' nested fields. This will be useful when we export/import Object Entries with permissions through Batch.