A simple solution that is easy to implement, tamper-proof and future-ready.
In 2023, the Internet started buzzing with AI agents, crawlers and language models all vying for their moment of glory.
We've learned that they can be clever, creative, productive and resourceful
As with all new technologies, new challenges also emerged.
Too often they are low quality, out of touch, lack integrity and attribution.
Content creators need a way to verify and demonstrate ownership, receive attribution even negotiate automatic compensation.
Teams building AI struggle to know what content is fair use, whether it has permissive licensing and so on.
We consider the following challenges with respect to AI so that society benefits as a whole.
2.1. Ensure Trustworthiness: Ensure that the facts are trustworthy and immutable - fostering trust among stakeholders.
2.2. Semantic Interoperability: Incorporating standards such as PROV-O, SKOS, and VOID annotations facilitates semantic interoperability. Enable seamless information exchange and collaboration across diverse domains and platforms.
2.3. Knowledge Curation: The fact web should serve as a structured repository for organizing and discovering relevant facts by adhering to standards, ontologies and vocabularies, such as PROV-O, SKOS, and VOID.
2.4. Transparent and Reproducible: By capturing provenance information using PROV-O, the fact web enables transparent and reproducible research. Auditors can trace the lineage of data and assertions, understand how they were derived or obtained, and verify their authenticity.
2.5. Reasoning and Analysis: The fact web facilitates automated reasoning and analysis. By representing data and relationships using standardized RDF, tools can infer new knowledge, detect patterns, and derive insights.
2.6. Privacy and Security: While promoting openness and transparency maintain privacy and security of sensitive information. Access control is part of the graph, identifying confidential data, authorized agents, privacy regulations and ethical standards.
2.7. Collaboration and Sharing: By providing APIs, query interfaces, and visualization tools, the graph enables seamless collaboration and communication, accelerating the pace of discovery and innovation.
In our architecture, a fact is a small atomic unit of knowledge - serialized in JSON-LD format.
The web links one fact to another - in same or separate documents - a technique called linked data.
We consider a grounded fact to be one stored on IPFS as an immutable record.
At runtime, URLs within the JSON-LD may be retrieved at the agent's discretion.
Unlike IPFS, Internet facts are not tamper-proof. They are dynamic so facts may be ingested by the agent, as required.
The technical architecture for fact claims consists of several key components:
With Linked Data, concepts and documents can be inter-related to describe almost anything.
graph TB;
Organization["schema:Organization"] -->|schema:offers| Services["schema:Service"]
Organization2 -->|my:supplier| Organization["schema:Organization"]
Products["schema:Product"] -->|schema:hasPart| Outcomes["my:Outcomes"]
Services -->|schema:hasPart| Outcomes
Products -->|my:supplier| SupplyChain["schema:Product"]
Services -->|my:provider| Organization2
Organization -->|my:partners| Organization2["schema:Organization"]
| Solution Domain | Use Cases |
|---|---|
| Ecommerce | Promote with Schema.org - Products, Events, Services, Offers, Loans/Credit |
| Content Creators | Employ metadata standards like Dublin Core and IPTC for multimedia curation - Creative Commons for licensing |
| Education | Ensure consistency through IMS Global Learning and SCORM - Learning Resource Metadata Initiative (LRMI) and CiTo (Citations) |
| Regulatory Compliance | Verifiable reports with XBRL, FIBO, and FIGREGONT - Ensure transparency and traceability using standards like PROV-O |
| Financial Auditing | Financial data exchange with XBRL - data auditability with PROV-O |
| Supply Chain Management | Traceability in supply chain processes with GS1 Standards - Utilize PROV-O for tracking changes |
| Healthcare Data Management | Standards like HL7 and FHIR for semantic healthcare data |
| Intellectual Property | Define digital rights with W3C ODRL for IP management |
| Research & Innovation | Track research activities with PROV-O and SKOS - citeusing CiTo (Citations) |
| Environmental Sustainability | Model environmental data using OGC SOSA/SSN |
| Legal Contracts and Agreements | Model legal matter Legal Core Ontology - Define digital rights using W3C ODRL |
| Identity and Access Management | Digital credentials with W3C VC and DID |
| Energy Trading and Grid Management | Energy market information with IEC CIM |
| Credential Verification | Credentials using W3C VC and Open Badges |
| Asset Tokenization and Management | Define tokenomics with ERC-20 and ERC-721 |
| News and Data-driven Narratives | Story-driven data exchange with NewsML-G2 and NITF |