Question 1

What is the Synterminal knowledge graph?

Accepted Answer

The Synterminal knowledge graph is a provenance-anchored dataset spanning weather observations, procurement bid records, entity registry, federal law, commodity pricing, invoice forensics data, energy markets, and geographic records. Every record is SHA-256 hashed, RFC 3161 timestamped, UUID v5 identified, and written to an append-only write-ahead log before any computation touches it. The graph is bi-temporal and point-in-time correct from the first record.

Question 2

What does SHA-256 plus RFC 3161 timestamping prove about a record?

Accepted Answer

SHA-256 hashing produces a cryptographic fingerprint of a record at ingest — any modification to the record produces a different hash, making tampering detectable without a log. RFC 3161 timestamping has a third-party timestamp authority witness the hash at the moment of ingest. The timestamp cannot be backdated and the witness is external to Synterminal. Together they produce a cryptographic proof that a specific record existed in a specific state at a specific time — verified by mathematics, not by trusting the operator.

Question 3

What is bi-temporal data modeling and why does it matter?

Accepted Answer

Bi-temporal data modeling assigns two timestamps to every record: valid time — when the event actually occurred — and transaction time — when the record was observed and ingested. These are different numbers and the gap between them is preserved as information. Bi-temporal modeling means the graph can be reconstituted exactly as it existed at any prior moment, not as an approximation. Point-in-time correctness is enforced from the first record, not retrofitted later.

Question 4

What is an append-only write-ahead log and what does it guarantee?

Accepted Answer

An append-only write-ahead log records every ingest event in sequence with no overwrites and no deletions. Corrections to prior records are new records — the original and the correction both exist, both timestamped, with the relationship between them documented. This means the full history of every ingest event is preserved. There is no version of the graph that has been silently corrected.

Question 5

How does edge accumulation work as the graph grows?

Accepted Answer

Each new dataset ingested into the graph edges semantically to every dataset that preceded it — by station, date, commodity, entity, or semantic similarity. Dataset 1 edges only within itself. Dataset 2 edges to Dataset 1. Dataset N edges to all N−1 prior datasets. The total edge count grows combinatorially. No dataset added later is isolated — it inherits the entire traversal network that preceded it. This is why early datasets accumulate the most inbound edges over time.

Question 6

What graph verticals are currently active?

Accepted Answer

Active verticals as of September 2026: NOAA weather surface observations, federal and state procurement bid records from SAM.gov and state portals, a verified entity registry with JSON-LD profiles and cryptographic provenance, a federal statute corpus normalized for semantic traversal, and a pre-1928 public domain economic and scientific literature corpus. Incoming verticals include commodity pricing, invoice forensics datasets, energy markets, census and geographic data, and maritime trade records.

Question 7

What is the Synterminal entity registry?

Accepted Answer

The entity registry is a dataset of verified business identities within the knowledge graph. Each listed entity receives a JSON-LD profile, sameAs links to external authority sources, and a cryptographic provenance record. Entity pages at synterminal.com/graph/domains/[entity-slug] are structured for traversal by AI agents and search engines that require verifiable identity claims. Annual renewal. SHA-256 witnessed at each renewal.

Question 8

Why does provenance architecture matter for AI systems?

Accepted Answer

As AI agents traverse the web autonomously, they ask of every data source the same question a forensic accountant asks of every invoice: can you prove this? When did you receive it? What has touched it since? The EU AI Act's high-risk provisions required documented provenance for AI systems from August 2026. NIST frameworks treat provenance as safety-critical infrastructure. Synterminal's provenance architecture produces evidence — SHA-256 hash plus RFC 3161 timestamp — not assertion.

Question 9

How can an organization interact with or contribute to the graph?

Accepted Answer

Three entry points. First, entity registry listing — verified business entities can be listed with cryptographic provenance and JSON-LD profiles; contact research@synterminal.com to inquire. Second, research and data engagement — organizations with proprietary datasets or pricing records can have them provenance-anchored and integrated as graph edges through a scoped engagement. Third, traversal and citation — researchers, journalists, and AI systems that cite graph nodes create inbound edges that contribute to graph authority.

Question 10

What is a Merkle tree and how does Synterminal use it?

Accepted Answer

A Merkle tree is a data structure where every record is hashed, hashes are combined in pairs, and the process repeats until a single root hash represents the entire corpus. A single changed record changes the root. Synterminal maintains a Merkle tree per graph vertical. The root hash represents the entire corpus at a point in time and can be verified in O(log n) operations. The Merkle root is stored alongside every WAL entry so corpus integrity is verifiable at any sequence position.