Defining schemas for RDF graphs

This session is about expressing syntactic constraints on RDF graph using the Shapes Constraint Language SHACL.

Objectives

• Understand existing shape schemas.
• Define constraints on what graph shapes are expected in the case of product data.
• Validate your shapes against RDF data.

Existing shapes for schema.org

TopQuadrant, a software company specialised in Semantic Web data management, provides a shape schema that expresses some of the constraints described informally in Schema.org specifications.

Answer the following questions in the email you send at the end of the session to Antoine Zimmermann and Victor Charpenay.

• When was this version of the SHACL schema of schema.org edited?
• Find where the schema:Order class is defined. What is the datatype for the value of the rdfs:comment for schema:Order?
• You should see that there are lots of values for sh:property. Consider schema:Order-orderDate. Find its definition in the file. What is the type of this entity?
• There is a constraint defined for schema:Order-orderDate as an sh:or constraint. Assuming ex:order is an instance of schema:Order, to which kind of entities does this constraint apply?. What is the constraint about? Guess what the constraint is telling by providing a correct example and an incorrect example. Write your answer in the email to us.
• Using the SHACL playground, you can test if an RDF graph validates a SHACL document. The box on the top left is where you enter the shape document. You can copy-paste the document from TopQuadrant. The box on the top right is where you put the RDF graph for the things of interest. For example, describe the order ex:order using schema.org, assming it has been placed today on LDLC for a Seagate BarraCuda hard disk drive at the price indicated in the offer. Include a description of the product and the offer as well (hint: the product and the offer are already described in JSON-LD in the web page). Refer to the documentation for schema:Order to describe the order. Then validate your entire graph against the SHACL schema. Copy the input RDF graph and the validation report to your email to us (both graphs should be in Turtle).

Write your own shape constraints

We will add constraints to orders such that every order must be associated with a date.

• Update the SHACL schema from TopQuadrant such that the property shape for schema:orderDate has exactly 1 date. This can be done using sh:minCount and sh:maxCount.
• You can say that the Order shape has either: (1) a date attached to it, or (2) that its associated offer has itself a date. Use sh:or to provide alternative constraint for orders that model this case.
• Create a new shape for LDLC orders. We want to add the constraint that the seller is LDLC and that the order number is a string with a specific format defined by LDLC: ldlc-[some number].
• You can express some SHACL constraints as SPARQL queries. Write a SPARQL query returning all nodes violating the following constraint: if the order date is of type xsd:dateTime, it must be between 9AM and 7PM. For that, you will need some of the SPARQL built-in functions.
• Conversely, you can express SPARQL queries as SHACL constraints. Go back to the instructions for the SPARQL session. Among all queries in that session, choose one and reformulate it as a SHACL shape with constraints.