Indexing-time process in Azure AI Search where skills such as OCR, key phrase extraction, entity recognition and image analysis turn raw content into fields that can be searched.
Also called skillset.
Read more: Microsoft Learn
In the Ultra Transcenders books
Each book explains AI enrichment in context, with comparison tables and the common traps.
Terms in this definition
- Azure AI Search
Managed Azure service that builds indexes over your content and answers keyword, vector, hybrid and semantically ranked queries, optionally with AI enrichment. Retrieval-augmented generation, Foundry agents and knowledge mining all use it to fetch relevant content.
- WHERE
Limits a SELECT, UPDATE or DELETE to just the rows meeting a condition. Omit it, and the statement hits every row.
- OCR
Optical character recognition. It reads handwritten or printed text from documents and images, making it available for analysis, something Azure Language cannot do with scanned images by itself.
- Key phrase extraction
Capability of Azure Language that pulls out the main ideas in a piece of text and returns them as a simple list, without confidence scores or categories.
- Entity
Something like a product or customer that an organisation stores data about. Its characteristics are its attributes, and every individual record is one instance.
- Image Analysis
Azure Vision API that returns several visual features in a single request: for instance tags, a caption, text recognised by Read OCR and the bounding boxes of detected objects.
- TURN
If a direct link can't be made, RDP Shortpath for Windows 365 relays UDP traffic via Microsoft servers on port 3478 instead.
Related terms
- Azure OpenAI Embedding skill
Skill in an AI Search skillset that sends each chunk to an embedding model deployment (text-embedding-3-small, text-embedding-3-large or ada-002) during indexing to produce vectors; charges come from Azure OpenAI.
- Billable Foundry resource (skillset)
Key or identity of a Foundry resource linked to a skillset, letting built-in Vision, Language and Translator skills process more than the free allowance of 20 documents per indexer per day; a single multi-service resource is enough for all of them.
- Field mappings
Indexer configuration passing a source field unchanged into an index field with another name or type. It is applied once documents are cracked and before any skillset runs.
- Indexer
The pull-model crawler in Azure AI Search that takes data from a single data source, can apply a skillset and loads the output into a single index. It runs on demand or on a schedule, at most every five minutes.
- Knowledge store
Property of an Azure AI Search skillset that uses projections to save enriched output into Azure Storage tables and blobs, so it can be analysed outside the search index.
- Output field mappings
Skills create nodes in the enriched document; this indexer setting routes them into index fields. It runs after the skillset, and enriched content is only indexed when mapped here.
- Parallel indexing
Runs several indexers at once, one per partition of the content (a container or virtual folder), each with a dedicated data source but sharing one skillset and target index. Concurrency is capped at about one indexer job per search unit.