RDF graph querying using prefix names with SPARQL

Question

I'm currently creating an ontology for a project that deals with sensor data. I successfully created an RDF graph using RDFLib in Python. And then stored the graph using Jena TDB database. Now I want to query the database using SPARQL but I honestly don't know where to start.

For example if I m using already a knowing vocabulary lets say Friend of a Friend (FOAF) then I will simply execute a query something like this and I will get what I need:

SELECT ?name
WHERE {
    ?person foaf:name ?name .
}

But since I'm creating my RDF using different ontologies (a sensor ontologies) my aim is to execute this query and get all the data related:

SELECT *  {?dev ?lon ?lat}

where dev is the device name , lon is the longitude where the device is located and lat is the latitude. All this data are provided within the graph but its not yet defined to be queried.

To summarize, I'm trying to figure out a prefixed name for classes and so on, for example the class device can be represented by (dev, device...)

Can assigning labels in RDFLib be the solution or do I have to develop a vocabulary for querying ?

In general, you need to understand the dataset you're querying. It is conforms to an Ontology, simply use predicates and classes from that Ontology into your queries. For instance, assuming that there is a predicate to get the lightness value with URI http://something.org/sensor/hasLightnessValue, you can get the lightness value for your device that has a URI http://something.org/device/myDevice as <http://something.org/device/myDevice> <http://something.org/sensor/hasLightnessValue> ?value prefixes are just to make the query more readable for humans — William Kinaan
You really don't need to know anything about the Ontology to explore and understand the dataset. Use SPARQL, if just ?s ?p ?o queries with a LIMIT, to find what data exists and find how the properties, classes, etc. are defined and used. — scotthenninger

scotthenninger scotthenninger · Accepted Answer · 2016-04-19T17:51:38

RDF triples are designed to represent data with a subject, object, predicate structure. For example, the following is an RDF triple:

<http://example.org/FalaG> foaf:name "FalaG"^^xsd:string

...where foaf:name is a prefixed qname that expands to <http://xmlns.com/foaf/0.1/name>.

SPARQL is basically the same as the above triple (this is Turtle text serialization), except that you can introduce variables prefixed by a ?. SPARQL will then find matches in the data and bind the variables for each match.

The name of the variable has no inherent meaning other than as a variable name. So the following would find the above triple, amongst others that match:

SELECT ?p 
WHERE {
    ?p foaf:name ?o
}

Note I've purposefully put in nonsense variable names to outline that there is no inherent meaning to the variable names.

In terms of finding the URIs, you can use SPARQL to discover what values are found in the repository. Building on the previous query, you can find whatever other information is in the repository for FalaG with the following:

SELECT ?person ?p ?o
WHERE {
    ?person foaf:name "FalaG"^^xsd:string .
    ?person ?p ?o
}

To discover URIs for other properties, etc, use the following:

SELECT ?s ?p ?o
WHERE {
    ?s ?p ?o
} LIMIT

The LIMIT is important here, because any triple pattern with three variables is requesting all triples, which will likely timeout or crash the server. But by setting the limit, you can find how the data is modeled, then substitute values in ?s, '?pand?o`, etc. to look in more detail at specific parts of the data.

And yea, there are lots of sources for learning more about RDF and SPARQL.

RDF graph querying using prefix names with SPARQL

1 Answers