Viva Learning in-app searches are powered by Azure AI search.
Viva Learning semantically understands the content based on Title, Description,
Skill Tags, Keywords attribute of the content. These attributes are ingested
into Viva Learning either from out-of-box connectors like LMS and 3P connectors or pushed into Viva Learning using VLGraph APIs. 

When a user types a query in Viva Learning search, our
semantic engine understands the user intent and matches it with the content processed
previously. The semantic understanding happens for every user-query separately,
the process is optimized so that users do not see any delay in Search results. 

As Copilot in Viva Learning uses Viva Learning Search
internally, Viva Learning Copilot results and suggestions are also enhanced to
show intent based results. Semantic Search also takes the
“prioritized-provider” configuration into consideration while showing results
for every query.  

Permissions

Semantic search takes all existing permissions (permissions
set in Viva Learning, SharePoint or in LMS if applicable) into consideration
while showing results.

Impacted Surfaces

The following are surfaces are impacted by Semantic Search:

• Viva Learning Search engine results page
• Viva Learning Copilot
• Content Curation Modals in
• Learning Paths
• Manage Academies
• Manage Permissions
• Feature Set
• Learning Collections
• Learning Tab
• Compose extension in Teams Chat

Language Support

Semantic Search is supported in all languages
supported by Viva Learning.

Responsible AI: Please find Azure AI Search transparency note here.

The Microsoft Search team handles Viva Learning results in
Microsoft 365 search surfaces (such as SharePoint and Bing@Work). Therefore,
in-app Semantic Search results may differ from those in non-Viva Learning
surfaces.

[What you need to do to prepare:]

Since Semantic Search relies on content titles,
descriptions, skill tags, and keywords, Viva Learning recommends that learning
curators enrich their content with detailed information on these attributes.
This will help our semantic engine better understand the content and provide
more accurate suggestions for user queries.