§ 5@vector embeddings

Vector embeddings sit awkwardly in classical JSON-LD: they're not IRIs, not literals, and they explode RDF graphs. The @vector container preserves them through expansion and validates dimensionality at parse time.

Input

expected @dimensions

vector a

vector b

Output

{
  "termDefinition": {
    "embedding": {
      "@id": "https://example.org/embedding",
      "@container": "@vector",
      "@dimensions": 4
    }
  },
  "document": {
    "@context": {
      "embedding": {
        "@container": "@vector",
        "@dimensions": 4
      }
    },
    "@type": "Document",
    "title": "FLAIRS-39 paper",
    "embedding": [
      0.9,
      0.1,
      0.4,
      0.2
    ]
  },
  "validateA": {
    "valid": true,
    "errors": []
  },
  "validateB": {
    "valid": true,
    "errors": []
  },
  "cosineSimilarity": 0.9940023209537981
}