Introduction

Using Ontologies as Datastore and representing Data in Ontologies is one thing but at some point some might want to ask specific Questions about the Data and retrieve a result. I found after some research that this is not as simple as it might sound.

If anything here is not correct, please help me to improve and leave a comment.

Q1

Which stones are in a Silver Bucket?

Q2

How Many Stones are there weighting more than 20?

SQL

SQL is specialised to query data from the relational model. There are a lot of extensions and various different implementations.

From the described Bucket.owl Ontology the SQL Schema should look more or less like the following.

CREATE TABLE buckets(
    id INT PRIMARY KEY
,    material VARCHAR(40)
,     engraving VARCHAR(40) ARRAY[20]
);
 
CREATE TABLE stones(
    id INT PRIMARY KEY
,    weight INT NOT NULL
,    datefound DATETIME NOT NULL
,    in_bucket INT REFERENCES buckets(id)
);

Q1

SELECT s.*
FROM stones s JOIN buckets b ON b.id = s.id
WHERE b.material = 'Silver';

Q2

SELECT COUNT(*) 
FROM stones s
WHERE s.weight > 20;

SPARQL

Sparql is a query language for RDF. It has thus only limited Knowledge of the special semantics in OWL [2].

Q1

PREFIX buck: <http://www.yoshtec.com/ontology/test/Bucket#>
PREFIX xsd:    <http://www.w3.org/2001/XMLSchema#> 

SELECT  ?x 
WHERE { ?x buck:weight ?size 
     FILTER( ?size < "20"^^xsd:int )
    }

Q2

I have no Idea how to express this query in Sparql itself.

SWRL

SWRL is not really a query language, it is a Rule language! However it can be extended to handle the execution of queries. Protege has an extension Ontology which does just that: SQWRL. It consists mainly of a collection of buildin-Functions that resemble the Query features.

But there are extreme Drawbacks: currently the only Engine capable of processing the extensions is Jess [1]. But it is not general avialable under an open source license, so it might hinder people using it.

Q1

Bucket(?b)  ∧ Contains(?b, ?s)  ∧ Stone(?s)  ∧ Material(?b, ?m)  ∧ swrlb:equal(?m, "Silver") → sqwrl:select(?s)

Q2

Stone(?s)  ∧  weight(?s, ?w)  ∧  swrlb:greaterThan(?w, 20) → sqwrl:count(?s)

Further Reading

• A recent Presentation from Martin O'Connor: Efficiently Querying Relational Databases using OWL and SWRL