Utilizing Graph connectors simplifies the process of incorporating your data into Microsoft 365 search and Microsoft 365 Copilot, enhancing your experience with seamless integration. A fresh addition to Copilot's capabilities is the Relevance Tuning feature, that significantly enhances Copilot's functionality..
The Relevance Tuning feature allows for fine-tuning the relevance of search results, ensuring that Copilot delivers even more accurate and tailored responses. By leveraging this innovation, users can expect a marked improvement in the quality and precision of Copilot's outputs. Administrators now have the power to assign importance weights to different properties within Graph connections. This customization means that search results can be fine-tuned to align closely with an organization's unique requirements, making the discovery of relevant content more intuitive and efficient. Administrators can find out more about that topic at Microsoft Graph connectors overview for Microsoft Search.
In the Microsoft 365 admin portal, in the Search & intelligence menu, the existing Graph connectors can be found in the Customization menu under Relevance tuning, as here. We can select a connector, and click on the Configure rank hints link.
In the side panel, the relevance tuning can be done on the field level, as here. The Importance slider is depending on the type of the field and the properties (see Search attributes). The importance weights controls the relevance, from Default to Medium, High, and Very high. In this sample, the AccountName gets a higher relevance than the Title, and the Description.
This feature enhances the overall search experience by leveraging AI-powered models. The integration of Graph connector results into the search UI is smooth and seamless, providing users with a consistent and unified interface. Administrators can employ to adjust the relevance of search results according to their preferences.
Relevance Tuning is general available since May 2024 for all M365 tenants. No action is required from the administrators. Overall, this new feature empowers organizations to optimize their search capabilities within Microsoft 365, making it a cool and valuable addition.
See more details at Register and update schema for the Microsoft Graph connection and Search Relevance for Microsoft Graph Connectors - A look under the hood. These articles also explain the property attributes like Searchable, Queryable, Retrievable, and Refinable, Semantic labels, Relevance, and the new Rank hints. If you are a developer, check out Build 2024: What's new for Microsoft Graph.
Try improving the Copilot results of your own imported data with the Relevance Tuning feature of the Graph Connector!