Semantic Search

Endpoint

POST /api/v1/knowledge/search

Usage

Semantic search uses 3072-dimensional Gemini embeddings with cosine similarity. Use full sentences — it understands meaning, not just keywords.

curl -X POST -H "X-API-Key: $KEY" \
  -H "Content-Type: application/json" \
  "https://api.marc27.com/api/v1/knowledge/search" \
  -d '{"query": "materials that resist creep at high temperature for turbine applications", "limit": 5}'

Response

[
  {
    "doc_id": "Ni-based Superalloy",
    "content": "Topic: Ni-based Superalloy. Type: TOP",
    "similarity": 0.752,
    "metadata": {
      "entity_type": "TOP",
      "source": "embed-graph"
    }
  }
]

When to Use

Use Case	Method
Exact entity name	Graph Search (?q=Inconel+718)
Natural language question	Semantic Search
Multi-concept exploration	Research Mode
Specific property lookup	Property Search

Embedding

Store your own text in the vector database:

curl -X POST -H "X-API-Key: $KEY" \
  -H "Content-Type: application/json" \
  "https://api.marc27.com/api/v1/knowledge/embed" \
  -d '{"doc_id": "my-paper-1", "content": "Abstract text here..."}'