Skip to main content

Introduction to the Semantic search Documents API

  • April 5, 2024
  • 1 reply
  • 74 views
Lars Moastuen
Seasoned Practitioner

Semantic search is a new service under the existing Documents API in Cognite Data Fusion. For details on what semantic search is and an introduction to the service, please see attached presentation below.

The service lets you search for relevant parts (passages) of up to 100 documents by using advanced filters and semantic search queries. Semantic search works by comparing the semantic meaning of the search query to the semantic meaning of the document passages. The document passages are automatically derived and indexed upon file uploads (PDF only).

The flow for using semantic search on set of documents is as follows: 

  1. Upload the PDF documents to Cognite Data Fusion
  2. Query /documents/status to check if the document is ready 
  3. Query /documents/semantic/search to find relevant passages in the document(s)

Documentation for the service is being developed. Please find documentation for the respective endpoints below:

1 Attachments
Lars Moastuen (Software Engineer, Atlas AI)
    C
    A
    Carol Delavalli

    1 reply

    Carol Delavalli
    Active

    This is a great solution, which has great performance. Here are some suggestions for improvements:

    • The analysis has a limited number of files and would not be able to handle a large database.
    • Dealing only with PDF files also limits the basis for searching for information.
    • And, despite internally using a similarity index, the API does not return this in the response; It would be worth it to have that return.
    Terms & ConditionsCookie SettingsAccessibility statement