Queries that curate, verify and infer facts

With SPARQL, you can craft queries to retrieve the facts based on complex criteria.

The WHERE clause in a SPARQL query defined a pattern that matches statements { subject, tag and object }.

A semi-colon ; indicates a set of tags belong to same subject.

A period . indicates a closing statement, afterwards we need a new subject.

And the FILTER only includes statements that match a boolean expression.

Retrieve claims and their source:

SELECT ?claim ?by
WHERE {
  ?claim rdf:type fact:Claim ;
    prov:generatedBy ?by .
}

Retrieve claims about the SmartTrust topic:

SELECT ?claim
WHERE {
  ?claim skos:member self:SmartTrust .
}

Retrieve claims notarized by a smart contract:

SELECT ?claim
WHERE {
  ?claim rdf:type fact:Claim ;
         prov:wasGeneratedBy ?by .
  FILTER (startswith(str(?by), "ethereum://"))
}

Retrieve all organizations:

SELECT DISTINCT ?claim
WHERE {
  ?claim rdf:type schema:Organization.
}

Retrieve all grounded fact claims, including sameAs aliases:

SELECT ?claim ?by
WHERE {
  {
    ?claim (prov:wasGeneratedBy*/^prov:wasGeneratedBy) ?by .
  }
  UNION
  {
    ?alias owl:sameAs ?claim.
    ?claim (prov:wasGeneratedBy*/^prov:wasGeneratedBy) ?by .
  }
  FILTER (startswith(str(?by), "ipfs://") || startswith(str(?by), "ethereum://"))
}

Infer fact:Grounded:

INSERT {
  ?claim a fact:Grounded .
}
WHERE {
  {
    ?claim (prov:wasGeneratedBy*/^prov:wasGeneratedBy) ?by .
  }
  UNION
  {
    ?alias owl:sameAs ?claim .
    ?claim (prov:wasGeneratedBy*/^prov:wasGeneratedBy) ?by .
  }
  FILTER (startswith(str(?by), "ipfs://") || startswith(str(?by), "ethereum://"))
}

• The INSERT keyword indicates that you want to insert triples into the graph.
• The pattern in the WHERE clause remains the same as before, retrieving either the original claims or their aliases along with their corresponding sources.
• After the insertion, each matched ?claim will be tagged a fact:Grounded.

Mini SPARQL Tutorial

The WHERE clause in SPARQL serves as a pattern matching mechanism, allowing you to specify patterns that RDF triples must conform to in order to match the query.

1. Basic Pattern Matching:
   • The basic form of pattern matching in SPARQL involves matching triples directly. For example, ?claim rdf:type fact:Claim matches triples where ?claim is of type fact:Claim.
2. Optional Matching:
   • Optional patterns are specified using the OPTIONAL keyword. They allow you to specify patterns that may or may not match. If an optional pattern doesn't match, the variables within it will be bound to null.
   • In your query, there's an attempt to use OPTIONAL to find aliases connected via the owl:sameAs property. However, it's not fully utilized due to a misunderstanding of its usage in this context.
3. Union of Patterns:
   • The UNION keyword allows you to combine multiple patterns together. Each pattern separated by UNION operates independently, and results from each are combined.
   • Our query utilizes UNION to combine patterns for finding both the original claim and its aliases (connected via owl:sameAs, but schema:sameAs is also common).
4. Filtering:
   • Filtering is used to constrain query results based on specified conditions. Filters are applied after the basic pattern matching.
   • Your query utilizes FILTER to restrict sources based on their prefixes (ipfs:// or ethereum://).
5. Property Path Expressions:
   • Property paths allow you to traverse RDF graphs efficiently using concise syntax. They specify sequences of properties and/or inverse properties to follow.
   • Our query follows paths like (prov:wasGeneratedBy*/^prov:wasGeneratedBy) to traverse through chains of prov:wasGeneratedBy relationships, including their inverses, an exhaustive search for claims.