Recipe
Key quote extractor from a document
Extract the most impactful, quotable sentences from any document using Meridian's semantic chunking and salience scoring.
Overview
This recipe walks through feeding a document into Meridian, splitting it into semantic chunks, scoring each sentence for quotability, and returning a ranked list of key quotes suitable for summaries, social posts, or presentations.
Prerequisites
- Meridian API key with inference scope
- Document uploaded to a Meridian workspace
- Node.js 18+ or Python 3.10+ runtime
Steps
- Ingest documentPOST /v1/documents with the file or URL. Meridian chunks and indexes automatically.
- Run quote extractionPOST /v1/extract/quotes with document_id and optional max_quotes parameter.
- Filter resultsEach quote returns a salience score (0-1). Filter by threshold to keep only high-impact excerpts.
- ExportQuotes include source page and paragraph offsets for attribution.
Sample response
{
"quotes": [
{
"text": "The only limit to our realization...",
"score": 0.94,
"page": 3,
"paragraph": 2
}
]
}