By Rada Vassil. July, 2024
Azure AI Search processes documents using something called a skillset.
Think of a skillset as a set of tools that work together to improve and organize information from documents.
When you have a bunch of raw data, like customer reviews, a skillset can be used to analyze and enrich this information.
Here's how it works:
1. Skillset: This is a collection of tools, or "skills," that process the data. For example, one skill might translate text, another might recognize text in images, and yet another might detect the sentiment of a review.
2. Enriched Document: As the skills process the data, they create an "enriched document." Initially, this is just raw data, but as each skill does its job, the document becomes more structured and detailed.
3. Enrichment Tree: This is like a map of all the changes and improvements made to the document. It shows how data is transformed and organized by each skill.
4. Field Mappings: Once the data is processed, it needs to be put into a format that can be easily searched or stored. Field mappings help decide what parts of the enriched document get transferred to the final search index or knowledge store.
5. Example: Imagine you have reviews about hotels. One skill might split long reviews into smaller chunks, another might detect the language, and yet another might analyze the sentiment of each chunk. These improvements are tracked in the enrichment tree and finally put into a format that's ready for searching or further analysis.
Overall, skillsets help turn raw data into useful, organized information that can be easily searched or stored.