Examples

Data Ingestion

Fundamentals of loading data from files, URLs, and JSON into the system.

I/O Module

Modes of Ingestion

Getting data IN is the first step. The SDK supports structured (JSON) and unstructured (Text/PDF/URL) data.

Unstructured

Files, URLs, Raw Text. The system automatically chunks and extracts entities.

Structured

JSON data. Preserves the schema and relationships better.

Unified

ingest_unified auto-detects the best method.

Code Example

# 1. Ingest a URL (Wikipedia)
# The system scrapes, cleans, and builds a KG from the page
client.ingestion.upload_url(
    url="https://en.wikipedia.org/wiki/Marathon",
    kg_name="KG_Companion_Demo",
    source_name="Wiki: Marathon"
)

# 2. Ingest Structured JSON
# Best for data that is already rigorous
training_data = {
    "week_1": "Run 1 mile",
    "week_2": "Run 2 miles"
}

client.ingestion.ingest_json(
    data=training_data,
    kg_name="KG_Companion_Demo",
    source_name="Training Plan",
    processing_mode="schema_aware"
)

Key Takeaways

Ingestion is asynchronous. These methods return a job_id.
Everything ingested ends up in a Knowledge Graph (specified by kg_name).
Use ingest_json when you have structured data to avoid parsing errors.