Extraction

Your schema.
The right fields.

Extraction models turn text and parsed documents into records. Define the fields or entity types you need, then inspect the values and their source references.

Input

Text or parsed document content, plus a schema or entity definition.

Output

Typed data with text spans, page references, or source-row references.

Evaluation

Field accuracy, schema validity, missing fields, and source correctness.

extract.schema

Schema Extract

Define the fields you need and retain the source for each returned value.

Inspect sample
Sample output
{
  "data": {
    "invoice_id": "INV-204",
    "total_usd": 480
  },
  "sources": {
    "invoice_id": {
      "page": 1,
      "text": "INV-204"
    },
    "total_usd": {
      "page": 1,
      "text": "USD 480.00"
    }
  }
}
extract.entities

Entity Extract

Identify entities by type and return their original text spans.

Inspect sample
Sample output
{
  "entities": [
    {
      "type": "city",
      "text": "Oslo",
      "start": 8,
      "end": 12
    },
    {
      "type": "organization",
      "text": "Northstar Logistics",
      "start": 17,
      "end": 36
    }
  ]
}
extract.tables

Table Extract

Normalize parsed table cells into records with source-row references.

Inspect sample
Sample output
{
  "page": 1,
  "rows": [
    {
      "Item": "Cable",
      "Qty": 4,
      "Price": 120,
      "source_row": 1
    }
  ]
}
Model availability

This research interface is not yet available through a public API. Explore the examples or contact us about your application.

Build what comes next.

Talk to Istren