ECAI Knowledge Encoding Format (EKEF v0.3)

Overview

This version of the ECAI Knowledge Encoding Format uses VRLP (Verifiable Recursive Length Prefix) encoding for serialization of the knowledge tuple. VRLP is deterministic, compact, and binary-friendly, making it ideal for elliptic curve knowledge encoding.

1. Knowledge Tuple (Raw Form)

Each knowledge entry is a 5-element ordered list:

["gravity", "is proportional to", "mass", "newtonian physics", 1712493296]

All fields are:

Strings (`utf-8`) except for timestamp (`uint64`)
Canonicalized (lowercased, stripped of extra whitespace)

2. Canonical VRLP Encoding

Serialize using a VRLP encoder:

from eth_rlp import encode as rlp_encode  # or pyrlp for standard RLP

def encode_vrlp(subject, predicate, obj, context, timestamp):
    items = [
        subject.encode("utf-8"),
        predicate.encode("utf-8"),
        obj.encode("utf-8"),
        context.encode("utf-8"),
        timestamp.to_bytes(8, "big")
    ]
    return rlp_encode(items)

This produces a single `bytes` blob that is suitable for cryptographic hashing and elliptic curve mapping.

3. Hashing the VRLP

Hash the VRLP-encoded blob using SHA-256:

import hashlib

def hash_knowledge_vrlp(vrlp_bytes):
    return hashlib.sha256(vrlp_bytes).digest()

4. Mapping to Elliptic Curve Point

from ecdsa import SECP256k1, ellipticcurve

curve = SECP256k1.curve

def map_to_curve(hash_bytes):
    x = int.from_bytes(hash_bytes, 'big') % curve.p()
    while True:
        try:
            y = curve.y_values(x)[0]
            return (x, y)
        except Exception:
            x = (x + 1) % curve.p()

5. Final Encoded Knowledge Point

curve: secp256k1
x: 0xabc123...
y: 0xdef456...

The original VRLP-encoded data can be included for verification.
Knowledge entries can be signed and published as NFTs or in Merkle trees.

6. Benefits of VRLP

Fully deterministic
Compatible with Ethereum and smart contract standards
More compact and binary-aligned than JSON
Easy to hash and map onto elliptic curves
Enables future compatibility with zk-SNARKs and L2 chains

7. Applications

Verifiable knowledge graphs
On-chain reasoning agents
Cross-subfield knowledge retrieval
Decentralized consensus on scientific facts

DamageBDD