Introduction and Overview Grant Weddell September 10, 2013 Data on - - PowerPoint PPT Presentation

Information Integration on the WEB with RDF, OWL and SPARQL

Introduction and Overview

Grant Weddell September 10, 2013

Data on the WEB

Consider the HTML associated with the URI http://en.wikipedia.org/wiki/Resource Description Framework To humans (with browser):

Data on the WEB

Consider the HTML associated with the URI http://en.wikipedia.org/wiki/Resource Description Framework To machines:

Data on the WEB

Consider the HTML associated with the URI http://en.wikipedia.org/wiki/Resource Description Framework To usefully integrate this information, machines must

• 1. understand natural languages, and
• 2. have domain specific understandings of the world.

RDF and OWL are a solution:

• 1. Add HTML that encodes data and metadata in the form of

subject/predicate/object triples.

• 2. Add two new WEB functions

Fetch : URI → HTML SelectRDF : HTML → HTML FilterRDF : HTML → HTML

Information Integration

SQL

• Conceptual Schema
• Source1

(RDB) · · · Sourcen (RDB) ⇒ SPARQL

• Ontology (OWL)
• URI1

(RDF) · · · URIn (RDF) Relational setting:

◮ Source mappings, option 1: local as view (LAV) ◮ Source mappings, option 2: global as view (GAV) ◮ Query evaluation: optimization/compilation then plan execution ◮ Metadata operates as constraints ◮ Closed world

Information Integration

SQL

• Conceptual Schema
• Source1

(RDB) · · · Sourcen (RDB) ⇒ SPARQL

• Ontology (OWL)
• URI1

(RDF) · · · URIn (RDF) WEB setting:

◮ Object identity: literal values + uniform resource identifiers (URIs) ◮ Data and metadata: resource description framework (RDF) ◮ RDF: collection of subject/predicate/object triples ◮ Integration of information: i SelectRDF(Fetch(URIi))

Information Integration

SQL

• Conceptual Schema
• Source1

(RDB) · · · Sourcen (RDB) ⇒ SPARQL

• Ontology (OWL)
• URI1

(RDF) · · · URIn (RDF) WEB setting (cont’d):

◮ Query evaluation: optimization/compilation then plan execution ◮ Metadata can infer new data ◮ Open world

Information Integration

Example: (data) (metadata) John/age/32 student/subclass/human John/type/student human/exists-property/age Mary/type/student age/range/integer age/type/functional-property Query: known humans and their ages select ?x, ?y where {?x type human. ?x age ?y} Result:

◮ With basic RDF entailment: { } ◮ With OWL 2 direct semantics entailment:

{{?x = John, ?y = 32}}

Information Integration

Example: (data) (metadata) John/age/32 student/subclass/human John/type/student human/exists-property/age Mary/type/student age/range/integer age/type/functional-property Query: known humans that have an age select ?x where {?x type concept-intersection(human, exists-property(age))} Relies on OWL 2’s ability to express complex concepts: the intersection of the set of humans and the set of things having an age property Result:

◮ With basic RDF entailment: { } ◮ With OWL 2 direct semantics entailment:

{{?x = John}, {?x = Mary}}

Overview

There are three main topics:

• 1. RDF storage engines
• 2. SPARQL query evaluation
• 3. Ontology-based data access (OBDA)

Will also be useful to review

◮ complexity theory, to understand the difficulty of query evaluation, ◮ first order logic, which underlies both the relational and WEB setting and ◮ description logics (DLs), in particular the dialect SHROIQ(D).

Topics not covered:

• 1. schema integration
• 2. fact extraction
• 3. inductive reasoning