WSML2Reasoner - API and User Guide

1  Introduction

4  WSML2Reasoner Query Extensions

Purpose of the SQL Interface Extension/WSML-Flight-A Query Language

The purpose of the SQL Interface extension required to accomodate the WSML-Flight-A query language is to provide an easy to use syntactic extension for querying ontologies and to facilitate the following main features:

• A SQL like syntax, which a large number of IT-Professionals are well accustomed with.
• Aggregates, which allow to collect additional useful data from the query result itself, in a post-processing step.
• Further post-processing of query results, e.g. selecting only part of the solution set to a resolved query.

This is mainly a use case driven extension of SQL originally required to accomodate the WSML-Flight-A query language. Additional motivations and details are given in [1]. Furthermore since this query language is layered on top of WSML-Flight, there are no compatability issues. There is no need to adjust existing ontologies, reasoning components, or other existing dependent software due to the architecture chosen.

Syntax and Usage

The allowed syntax for WSML-Flight-A queries closely corresponds to standard SQL syntax. For an exhaustive, in-depth description we again refer to [1], while [2] and [3] provide a concise overview of the SQL syntax. This section is only meant to serve as a short introduction.

A WSML-Flight-A query in general is of the following form:

	SELECT X1, ...,Xn
	FROM o
	WHERE Q

Whereby Q is an ordinary WSML-Flight query, X₁, ..., X_n are a subset of the variables appearing in Q, and o is the IRI of the ontology over which the query is placed. The WSML-Flight-A extension allows for SQL-like projection via the select expression in the usual SQL-like syntax and semantics. It explicitly supports the following constructs:

• COUNT(*), COUNT
• MIN, MAX, SUM, AVG
• SOME, EVERY
• VAR_POP, VAR_SAMP, STDDEV_POP, STDDEV_SAMP

With the exception of COUNT(*), all aggregate functions exclude NULL values, as in SQL. The type of the returned value for SUM is subject to deterministic widening to ensure lossless results. Numeric aggregates obviously only return values if and only if all bindings of a variable are also numeric. The return value type for COUNT is integer, for MIN, MAX and AVG it is the same type as the column, for SOME and EVERY it is boolean. For VAR_POP, VAR_SAMP, STDDEV_POP and STDDEV_SAMP statistical functions, the type is always double. Additionally the usual further SQL constructs are possible:

• GROUP BY expression
• HAVING expression
• ORDER BY orderExpression
• LIMIT <limit> [OFFSET <offset>]

Where orderExpression denotes the usual:

{ column number | column alias | select Expression } [ASC | DESC]

And LIMIT n m creates the result set for the SELECT statement first and then discards the first n rows (OFFSET) and returns the first m rows of the remaining result set (LIMIT).

Examples

Following are some short examples for the query expressions. The basic usage of the query extension is very simple. Queries are passed as ordinary String objects, and the WSMLQuery class is the only entry point. Results are returned in the expected WSML2Reasoner wrappers.

 
	String query = "<query>"
	WSMLQuery wqe = new WSMLQuery();
	Set<Map<Variable, Term>> r = wqe.executeQuery(query);

	SELECT ?x
	FROM _http://someontology
	WHERE ?x memberOf ?y"

	SELECT ?place, ?employee
	FROM _http://someontology
	WHERE ?employee[hasWorkingPlace hasValue ?place]
	ORDER BY 1, 2

	SELECT ?place, COUNT(?place)
	FROM _http://someontology
	WHERE ?employee[hasWorkingPlace hasValue ?place]
	GROUP BY ?place
	HAVING COUNT(?place) > 4

	SELECT ?place, COUNT(?place)
	FROM _http://someontology

	WHERE ?employee[hasWorkingPlace hasValue ?place]
	GROUP BY ?place
	HAVING COUNT(?place) > 4
	ORDER BY ?place DESC
	LIMIT 2