zk-SPARQL

Verifiable and anonymous personal datastore supporting SPARQL queries

presented at Internet Identity Workshop (IIW) on 17th November, 2022

Transcript

1. zk-SPARQL:
   verifiable and anonymous
   personal datastore supporting
   SPARQL queries
   November 15, 2022
   Internet Identity Workshop (IIW) 35
   Dan Yamamoto (Internet Initiative Japan Inc.)
   Joint work with:
   Yuji Suga (Internet Initiative Japan Inc.)
   Kazue Sako (Waseda University)
   Photo: CC BY 2.0 - Kanko* / Flickr https://www.flickr.com/photos/kankan/33346213/
   

   View Slide

2. Datastore
   1
   Data Consumer Datastore Data Source
   query
   data
   result
   consume
   data
   Data Source
   ...
   

   View Slide

3. Verifiable Datastore
   2
   Data Consumer Datastore Data Source
   query
   signed data
   result
   + signatures
   verify
   &
   consume
   signed data
   Data Source
   ...
   authenticity
   

   View Slide

4. Verifiable Personal Datastore
   3
   Data Consumer Datastore Data Source
   query
   signed data
   result
   + signatures
   verify
   &
   consume
   signed data
   Data Source
   ...
   authenticity
   privacy
   (signed data containing the query results
   must be fully disclosed for verification)
   

   View Slide

5. Verifiable & Anonymous Personal Datastore
   4
   Data Consumer Datastore Data Source
   query
   signed data
   result
   + proofs
   with
   selective disclosure
   verify
   &
   consume
   signed data
   Data Source
   ...
   authenticity
   privacy
   

   View Slide

6. Verifiable & Anonymous Personal Datastore
   5
   Data Consumer Datastore Data Source
   query
   signed data
   result
   + proofs
   with
   selective
   disclosure
   verify
   &
   consume
   signed data
   Data Source
   ...
   Issuer
   Issuer
   VC
   VC
   VP
   Verifier Holder
   zk-SPARQL
   

   View Slide

7. Example
   6
   Data Consumer Datastore Data Source
   query
   signed data
   result
   + proofs
   with
   selective
   disclosure
   verify
   &
   consume
   signed data
   Data Source
   ...
   { "type": "VerifiableCredential",
   "credentialSubject": {
   "id": "did:example:John",
   "type": "Person",
   "name": "John Smith",
   "homeLocation": "http://example.org/cityA",
   "birthDate": "2000-01-01",
   },
   "proof": ... }
   { "type": "VerifiableCredential",
   "credentialSubject": {
   "id": "http://example.org/cityA",
   "type": "Place",
   "name": "City A",
   "maximumAtendeeCapacity": 10000
   },
   "proof": ... }
   SELECT ?population
   WHERE {
   ?s a :Person .
   ?s :homeLocation ?place .
   ?place a :Place .
   ?place :maximumAttendeeCapacity ?population .
   }
   { "type": "VerifiablePresentation",
   "verifiableCredential": [
   { "type": "VerifiableCredential",
   "credentialSubject": {
   "id": "anoni:eX2hjL",
   "type": "Person",
   "homeLocation": "anoni:38CqTd" },
   { "type": "VerifiableCredential",
   "credentialSubject": {
   "id": "anoni:eX2hjL",
   "type": "Person",
   "maximumAtendeeCapacity": 10000 }
   ],
   "proof": ... }
   { "population": 10000 }
   

   View Slide

8. Veanpods
   7
   ◼ Verifiable & Anonymous Personal Datastore
   • stores signed RDF data (= JSON-LD VC)
   • can be queried using zk-SPARQL, an extension of SPARQL
   • responds minimal result with zero-knowledge proof
   • supports unlinkability
   • and predicate proofs (as future work)
   • repository: https://github.com/zkp-ld/veanpods
   (super experimental version)
   Photo: CC BY 2.0 - Kanko* / Flickr https://www.flickr.com/photos/kankan/33346213/
   

   View Slide

9. Structure
   8
   RDF store
   zk-SPARQL
   processor
   SPARQL UI
   for owner
   BBS+ signatures
   for RDF
   bulletproofs,
   zk-SNARKs, ...
   secure storage
   SPARQL UI
   for consumer
   store-binding
   consumer
   authorization
   = YASGUI
   use
   SPARQL.js
   use
   quadstore
   based on
   jsonld-signatures-bbs
   Data Consumer Data Source
   Veanpods
   owner
   

   View Slide

10. Future Work
    9
    ◼ create, update, delete function (CRUD)
    ◼ more than SELECT queries, i.e., ASK, CONSTRUCT, ...
    ◼ rich predicate proofs
    ⚫ require general-purpose zero-knowledge proofs,
    e.g., zk-SNARK, zk-STARK, bulletproofs
    ◼ zk-GraphQL, zk-GQL, zk-Cypher, ...
    ⚫ (maybe) possible by using existing translators (e.g., GraphQL-LD by Comunica)
    ◼ associate with Solid Pods (need to enhance my Solid knowledge)
    ◼ secure storage for signed data (VCs) and keys
    ◼ security & privacy formalization
    ◼ and more
    ⚫ revocation, PPID, issuer-hiding, mobile, ...
    

    View Slide