JSON-LD and Schema.org

Observing deployed JSON-LD

In this part, you simply observe existing examples on the Web. JSON-LD data can be embedded directly in Web pages inside a <script> element. To recognise JSON-LD code, the <script> element must have the attribute type="application/ld+json". Also, the presence of the key @context in JSON data suggests that it is JSON-LD.

• Go to this Web page about a music label and look at the source of the page. Find JSON-LD code and copy it to your answer file. What entity (or entities) is (are) described there?
• Go to this Web page about a CPU chip. Can you decide, using only JSON-LD data from this page, whether the product is available for sale or not? How?
• Go to this page about a dish by Jamie Oliver. According to the JSON-LD description, what type of entity is described here? Convert the RDF graph in the JSON-LD data to Turtle, using any means you want. Copy the Turtle representation of the graph in your answer file.
• The above examples feature the most commonly used Schema.org classes and properties but there are many more. In your answer file, give the number of classes and the number of properties that are defined by Schema.org.

Writing a context for existing JSON files

We will enhance existing JSON data by adding a @context. To test JSON-LD code, you can use the JSON-LD Playground.

• Go to the Open Movie Database and try the examples. Take your favourite film and get the JSON response from the service that describes it. Then, in a separate file, make a JSON-LD @context that will turn the JSON response into an RDF graph. You may use Schema.org to map the JSON keys to Schema.org properties, etc.
• Make sure the values that identify real world entities are IRIs and not just strings. Also, be careful to have appropriate datatypes for literals.

Reshaping JSON-LD files

In the last session, you uploaded Turtle files to a Linked Data platform. The platform supports content negotiation: you may now ask to return the uploaded data in JSON-LD, instead of Turtle. Send a GET request with the appropriate header so that the returned RDF graph has content type application/ld+json.

The JSON-LD file you obtained is in an expanded form. No objects are nested inside other objects. Write a JSON-LD context with as many definitions as possible, to turn the expanded JSON-LD file to a more compact variant. Make sure the two files (the expanded JSON-LD file and a compacted variant that includes a @context and one or more objects) are equivalent. Two JSON-LD files are equivalent if they encode the same RDF triples. Re-upload the compact variant to the Linked Data platform under another URI (to compare them).

Writing JSON-LD using Schema.org

Consider the data you wrote in Turtle during the second session. Rewrite it in JSON-LD and try to use Schema.org concepts as much as possible.