A simple demonstration using linked biometrics with Claim Deduction and Verifiable Credentials for age verification. The demo is a simulated vending machine that performs age verification before dispensing items. Try it yourself here.
Uses:
This age verification method can be partially privacy preserving because biometric data is embedded in a signed credential rather than being stored in some central database. The person getting their age verified doesn't need to reveal much personally identifiable information to the verifier other than their face and the fact that they are older than a certain threshold. The credential can leave out data points like name and gender.
This demo implements and uses a local DID method called "did:demo". The vending machine trusts age claims made by did:demo:age_root. The "secret" key for that DID is in ./src/helpers/didcache.js. Feel free to experiment with issuing your own credentials.
Credentials in this demo are irrevocable but could in practice link to a revocation registry.
The vending machine accepts credential presentations and verifies them against the face it sees in its camera before dispensing an item.
In real large-scale use, it may not be efficient for a single entity to issue every age credential. As a proposed solution, and as an excuse to show off Dock's Claim Deduction engine, this demo allows for delegation of authority by issuers.
Dock credentials are modeled as RDF which gives them meaning in the context of the semantic web. It's possible for a computer perform deductive reasoning over RDF graphs. Using Claim Deduction, the vending machine can perform limited generic deductive reasoning given a set of allowed logical axioms. Check out src/helpers/rules.js to see the axioms accepted by the vending machine.
did:demo:age_root does not issue age credentials directly. Rather, did:demo:age_root signs a separate credential delegating authority to another issuer. The age credentials issued by this demo are wrapped in VCDM presentations and bundled with the other credentials that comprise a delegation chain. A deductive proof is attached to the presentation. The proof is machine generated, and derives an implication relationship from truths made self-evident by the credentials to the fact in question: "Is the person in the image old enough?".
The proofs generated by this demo tend to take the form:
// age_root bequeaths delegation authority to delegate1
[did:demo:age_root type AgeDelegate] and
[did:demo:age_root claims [did:demo:delegate1 type AgeDelegate]]
therefore [did:demo:delegate1 type AgeDelegate]
// who in turn bequeaths delegation authority to delegate2
[did:demo:delegate1 type AgeDelegate] and
[did:demo:delegate1 claims [did:demo:delegate2 type AgeDelegate]]
therefore [did:demo:delegate2 type AgeDelegate]
// delegate2 claims the image pictures a single individual who is old enough
[did:demo:delegate2 type AgeDelegate] and
[did:demo:delegate3 claims [<image> type OfAge]]
therefore [<image> type OfAge]
The vending machine in this demo lays the burden of proof on the presenter, but it's possible to perform inference without requiring an embedded proof. As intuition would indicate, inference is computationally harder than proof verification.
The following is an example of a VCDM presentation emitted and accepted by this demo. Base 64 embedded images are hidden for readability.
{
"@context": ["https://www.w3.org/2018/credentials/v1"],
"type": ["VerifiablePresentation"],
"verifiableCredential": [
{
"@context": ["https://www.w3.org/2018/credentials/v1"],
"id": "uuid:67d8a4f9-61b7-4f98-a830-e7fef6fd274f",
"type": ["VerifiableCredential"],
"issuer": "did:demo:age_root",
"credentialSubject": {
"@id": "did:demo:delegate1",
"@type": "https://example.com/AgeDelegate"
},
"issuanceDate": "2020-10-12T23:59:06.320Z",
"proof": {
"type": "Ed25519Signature2018",
"created": "2020-10-12T23:59:06Z",
"jws": "eyJhbGciOiJFZERTQSIsImI2NCI6ZmFsc2UsImNyaXQiOlsiYjY0Il19..GjaPOrG-ca50NfE7zmx0Ff5jd2TaaNabCkh47gVj3lbYTxe5EkyIVP49BuAu9JNM1K5xB1V6XaH0lPBCkARfBw",
"proofPurpose": "assertionMethod",
"verificationMethod": "did:demo:age_root#keys-1"
}
},
{
"@context": ["https://www.w3.org/2018/credentials/v1"],
"id": "uuid:0f1da9fc-54d8-483c-ab32-8031ba380bd2",
"type": ["VerifiableCredential"],
"issuer": "did:demo:delegate1",
"credentialSubject": {
"@id": "did:demo:delegate2",
"@type": "https://example.com/AgeDelegate"
},
"issuanceDate": "2020-10-12T23:59:06.321Z",
"proof": {
"type": "Ed25519Signature2018",
"created": "2020-10-12T23:59:06Z",
"jws": "eyJhbGciOiJFZERTQSIsImI2NCI6ZmFsc2UsImNyaXQiOlsiYjY0Il19..sHGEIviELWLemtuc4n0rIXS9z90j3h8_N3Css3357y-bFZtzcPyagt7T8UHnVtminj4GZdxZTKp_GRnfMBHrAA",
"proofPurpose": "assertionMethod",
"verificationMethod": "did:demo:delegate1#keys-1"
}
},
{
"@context": ["https://www.w3.org/2018/credentials/v1"],
"id": "uuid:d8b2328b-ccf4-4bc9-b5e3-62c8ae58f13d",
"type": ["VerifiableCredential"],
"issuer": "did:demo:delegate2",
"credentialSubject": {
"@id": "data:image/png;base64,iVBORw0KG...etc...",
"@type": "https://example.com/OfAge"
},
"issuanceDate": "2020-10-12T23:59:06.322Z",
"proof": {
"type": "Ed25519Signature2018",
"created": "2020-10-12T23:59:06Z",
"jws": "eyJhbGciOiJFZERTQSIsImI2NCI6ZmFsc2UsImNyaXQiOlsiYjY0Il19..B8UjkhNXO5U-YWNxHZi-iJRoOIvEhfbESwzlGeDrd1ysuKwtb6eIptdrVJ675LdFkT4WNeCSsHzuTiKGMsnBDw",
"proofPurpose": "assertionMethod",
"verificationMethod": "did:demo:delegate2#keys-1"
}
}
],
"id": "uuid:71d67df1-419d-4382-af5e-5f240c9effec",
"holder": "uuid:e5dac0a1-06ec-4721-94ce-8dd0f3b3d821",
"https://www.dock.io/rdf2020#logicV1": {
"@type": "@json",
"@value": [
{
"rule_index": 0,
"instantiations": []
},
{
"rule_index": 1,
"instantiations": [
{"Iri": "did:demo:age_root"},
{"Blank": "_:b0"},
{"Iri": "did:demo:delegate1"}
]
},
{
"rule_index": 1,
"instantiations": [
{"Iri": "did:demo:delegate1"},
{"Blank": "_:b5"},
{"Iri": "did:demo:delegate2"}
]
},
{
"rule_index": 2,
"instantiations": [
{"Iri": "did:demo:delegate2"},
{"Blank": "_:b10"},
{"Iri": "data:image/png;base64,iVBORw0KG...etc..."}
]
}
]
}
}A standard camera as a facial recognition device can be tricked by holding a printed image of a face in front of your own. Use of specialized biometric scanning hardware may mitigate the problem. Suggestions are much appreciated; feel free to open a github issue.