Keyword Extraction helps you send text and get the most relevant keywords back. Use it when you need a quick way to summarize a document, index content, or power search and tagging workflows.
POST a body object with text, and optionally set language. If you want ApyHub to handle the request, leave requested_service at its default of apyhub; if you need to route through Azure, set it to azure and include the azure credentials object. The service is designed for plain text input, so it fits articles, support tickets, notes, and other unstructured content.
The response returns a data object with either an apyhub result or an azure result, depending on the service you request. That makes it useful for building content pipelines where you need keyword signals for search relevance, topic clustering, or lightweight document classification.
If you are normalising incoming text at scale, Keyword Extraction gives you a simple interface for turning raw paragraphs into structured keyword output without adding a separate NLP step to your app.
| Provider (requested_service) | Atoms |
|---|---|
| Azure | 500 |
| ApyHub | 2000 |
Call this endpoint with your ApyHub API key:
curl -X POST "http://localhost:8080/apyhub/extract-keywords-from-text" \
-H 'apy-token: $APY_TOKEN' \
-H 'Content-Type: application/json' \
-d '{"language":"en","requested_service":"apyhub","text":"This document describes details on security protocols and cryptography."}'Loading your default key…Extract keywords from text
Extracts keywords from the text you provide in the request body. You can optionally include a language hint, custom Azure credentials, and a requested_service value to choose between the Azure and ApyHub keyword-extraction responses.
Documentation imported from spec — please review and complete before submitting.
bad_requestRequired parameter missing or malformed body.
unauthorizedAPI key missing, revoked, or not authorized for this service.
rate_limitedYour plan's per-second rate exceeded. Retry with exponential backoff.
upstream_busyBackend temporarily unavailable. Try again in a few seconds.