# Information Extraction

## Overview

Fastino’s Information Extraction model is built to detect and extract structured data from unstructured text across diverse domains. 

The model operates in a zero-shot fashion, meaning users can define the specific entity types they want to extract at runtime without needing to retrain. It’s ideal for structuring data from documents, knowledge bases, chat transcripts, and narrative content. The model can also be fine-tuned on custom schemas for higher precision in domain-specific tasks.

***

## Example Use Cases

* Structuring key information from documents, news, or web pages
* Extracting facts and metrics from research papers or reports
* Real-time entity extraction from support chats, call transcripts, or emails
* Capturing medical, legal, or financial data for downstream processing

***

## Domains

Fastino’s Information Extraction model supports wide-ranging semantic labeling across many domains. It’s designed to help you extract structured information from unstructured text for use in knowledge bases, search, analytics, and downstream automation.

<table><thead><tr><th width="149.9423828125">Domain</th><th width="333.6337890625">Description</th><th>Example Entities</th></tr></thead><tbody><tr><td>People &#x26; Demographics</td><td>Extract identities, roles, and attributes related to individuals. Useful for profiling, biography parsing, or user interaction logs.</td><td>name, gender, age, nationality, occupation</td></tr><tr><td>Healthcare &#x26; Life Sciences</td><td>Capture medically relevant entities from clinical narratives, research literature, or patient records.</td><td>symptom, diagnosis, treatment, medication, medical condition, lab result</td></tr><tr><td>Places &#x26; Geography</td><td>Detect locations, addresses, and spatial markers in documents or conversations for mapping, logistics, and geoparsing.</td><td>street address, city, postal code, country, geo coordinates</td></tr><tr><td>Time &#x26; Events</td><td>Extract absolute and relative time references, event dates, durations, and cycles. Ideal for timeline construction or planning systems.</td><td>date, time, deadline, hospital admission date, flight date</td></tr><tr><td>Scientific &#x26; Academic</td><td>Identify structured elements in academic papers, research notes, or technical documentation.</td><td>theory, equation, dataset, gene, AI/ML model, publication year</td></tr><tr><td>Finance &#x26; Commerce</td><td>Extract data for transaction analytics, document structuring, or entity resolution in financial records and receipts.</td><td>transaction id, loan amount, account balance, stock ticker, revenue</td></tr><tr><td>Legal &#x26; Government</td><td>Detect references to laws, legal entities, identifiers, and procedures for case intake, compliance, or policy review.</td><td>legal case, passport number, visa type, regulation, contract type</td></tr><tr><td>Technology &#x26; Security</td><td>Extract identifiers and attributes related to digital systems, software metadata, and authentication artifacts.</td><td>ip address, api key, device id, commit hash, authentication token</td></tr><tr><td>Media &#x26; Entertainment</td><td>Parse structured references in media content, entertainment metadata, and online content streams.</td><td>movie, book, artist or performer, genre, song</td></tr><tr><td>Education &#x26; Skills</td><td>Extract academic background, credentials, and skill indicators from resumes, bios, or learning transcripts.</td><td>degree, university, graduation year, skill, GPA</td></tr></tbody></table>

***

### Usage

{% code title="Example Body" %}

```json
{
  "model_id": "fastino-data-structuring-••••••••••",
  "input": [
    {
      "text": "Albert Einstein proposed special relativity in 1905. This theory redefined concepts of space and time. E = mc² emerged from these groundbreaking ideas.",
      "parameters": {
        "entity_types": [
          "person",
          "theory",
          "date",
          "equation"
        ]
      }
    }
  ]
}
```

{% endcode %}

{% code title="Example Response" %}

```json
[
  {
    "input": "Albert Einstein proposed special relativity in 1905. This theory redefined concepts of space and time. E = mc² emerged from these groundbreaking ideas.",
    "latency_ms": 84.12,
    "message": "Responses from endpoint.",
    "output": {
      "entities": [
        {
          "start": 0,
          "end": 15,
          "label": "person",
          "text": "Albert Einstein",
          "score": 0.986
        },
        {
          "start": 25,
          "end": 45,
          "label": "theory",
          "text": "special relativity",
          "score": 0.934
        },
        {
          "start": 49,
          "end": 53,
          "label": "date",
          "text": "1905",
          "score": 0.921
        },
        {
          "start": 101,
          "end": 108,
          "label": "equation",
          "text": "E = mc²",
          "score": 0.947
        }
      ]
    },
    "status": "success"
  }
]
```

{% endcode %}


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://fastino-1.gitbook.io/docs/models/information-extraction.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.