Vector Search

Vectorization is the process of embedding your unstructured data into numerical representations of data that machines can read. Vector embeddings are used to measure the semantic closeness of different pieces of text to create accurate and relevant results in your RAG and agent queries.

Data 360 indexes this data. When you create a vector search index using the Data 360 search index UI or Connect API, you first chunk the unstructured data stored in DMOs and UDMOs. Chunking breaks the data into manageable, semantically meaningful units. Next Data Cloud generates vector embeddings from those chunks. Chunks are stored in chunk data model objects (CDMOs) and vector embeddings are stored in index daa model objects (IDMOs).

After the vector embeddings are added to the search index, you can perform vector searches on that data from apps such as Prompt Builder, AI agents, or Tableau.

• Create a Vector Search Index with Advanced Setup
  When you create a vector search index configuration for a data model object, Data 360 breaks up the referenced data into semantically related chunks and generates searchable vectors. Your RAG and agent workflows use these vectors to find and return semantically similar items when they run queries. To get more granular control over your search index configuration, use the advanced setup, which guides you through each step of the process.
• Run Vector Searches Using Query API
  Use the Query API to run vector searches in Data 360. Use these examples as starting points.
• Retrieving Content with Vector Search
  With vector search and AI agent queries, you can capture queries, search across structured and unstructured data sources, and receive summarized responses that improve efficiency and accuracy.
• Enhancing Data Analysis with Vector Search in Tableau
  Find and analyze data based on meanings, not just keywords, by using vector search in a Tableau query. Vector search broadens the scope of data exploration as well as enhances the accuracy and relevance of the insights by identifying semantic similarities rather than relying solely on exact keyword matches.