Download presentation
Presentation is loading. Please wait.
1
Dr. Jim Bowring Computer Science Department College of Charleston CSIS 690 (633) May Evening 2009 Semantic Web Principles and Practice Class 3: 18 May 2009
2
Class 3: Roadmap Announcements and Assignments Questions Info Overload responses FOAF responses SWWO Chapter 2 SWP Chapter 3
3
Semantic Modeling Models: abstraction --> communication provide explanations provide predictions mediate among viewpoints
4
Models for Human Communication Humans use informal models rely on context for interpretation legislation adjudication required Web examples: newsgroups, mailing lists FAQs categories of goods, etc. community tagging
5
Talmudic Model Layering achieved over time by incorporating interpretative sources authoritative sources Such informal models advance the collection of human knowledge aid communication require tuning and layering to resolve
6
Reasoning Inductive reasoning true premises mean likely true conclusions Analogical reasoning = simple induction specific objects, not general premises Deductive reasoning true premises guarantee true conclusions
7
Explanation and Prediction Explanation = trace of transformations from first principles via rules to a phenomenon Prediction = explanation Note: reading on semantic provenance Formal model provides objective way to state premises and rules What happens to the layers of an informal model when we model formally? SW: formal model with inferences
8
Mediating Variability Pluto example: 21st century, post 2006; astrological Solution: hierarchy of classes SSB astro:Planet horo:Planet IAU:Planet
9
What if … No clear class hierarchy? We merge models with conflicting information?
10
Expressivity in Models for SW RDF = resource description framework basic modeling for AAA, layering RDFS = RDF Schema language RDFS-Plus = subset of OWL relates properties OWL = Ontology Web Language adds logic for constraints between classes, entities and classes
11
More Models for SW DAML+OIL DAML+OIL = DARPA Agent Markup Language W3C RIF W3C RIF = Rule Interchange Format RuleML = Rule Markup Initiative R2ML R2ML = REWERSE I1 Rule Markup Language integrates OCL, SWRL, RuleML SWRL SWRL = Semantic Web Rule Language integrates OWL and RuleML
12
RDF SW uses primitive meaning of semantics = referential : symbols refer to things Resource = thing = entity Resource Description Framework
13
Distributing Data across Web Relational DBs mix strategies (via global DBMS): local tables local records local projections RDF strategy: local cells ==> global reference to row, col Triples: subject predicate object rowcolcell node directed edgenode
14
Merging Data from Multiple Sources Each triple is a graph Graphs are merged by equating nodes How? Note: SDOs: Service Data ObjectsService Data Objects Java data abstraction leveraging XML and providing results as directed graphs
15
Namespaces, URIs, Identity Answer: URIs Uniform Resource Identifiers URIs look just like URLs, but can be dereferenced Confused? http://www.w3.org/TR/uri-clarification/#confusion
16
Qnames To make readable in print, use qnames namespace:identifier namespace defined elsewhere URIs use CamelCase Standard namespaces: xsd: XML schema definition xmlns: XML namespaces rdf: http://www.w3.org/1999/02/22-rdf-syntax-ns#http://www.w3.org/1999/02/22-rdf-syntax-ns# rdfs: http://www.w3.org/2000/01/rdf-schema#http://www.w3.org/2000/01/rdf-schema# owl: http://www.w3.org/2002/07/owl#http://www.w3.org/2002/07/owl#
17
RDF Identifiers rdf:type rdf:Property basic typing system for predicates lit:Shakespeare lit:wrote lit:King Lear lit:wrote rdf:type rdf:property
18
Challenge: RDF and Tabular Data Represent RDB table losslessly as RDF graph: URI for the database each record as tableNameUniqueKey each column as tableName_Attribute ==> triples for each cell Additionally need to express types as membership row rdf:type table How many triples used?
19
Reification Make statement about another statement Strategy 1: add another attribute (RDB) or triple works well for adding info, but not for qualifying Strategy 2: explicit reification ns:stmtrdf:subject lit:Shakespeare; rdf:predicate lit: wrote; rdf: object lit:Hamlet. Note: we have not asserted underlying triple THEN: web:Wikipedia m:says ns:stmt
20
Serialization Alternatives N-Triples :. Notation 3 RDF (N3) (Berners-Lee): binds local qnames to global URIs subject predicate object; predicate object; predicate object. subject predicate object1, object2, object3. abbreviations: rdf:type ==> “a” RDF/XML : (for machines) http://www.w3.org/TR/rdf-syntax-grammar/
21
Blank Nodes and Ordered Info If we know something exists, but not its identity: Use existential qualifier by: [ rdf:type bio:Woman; bio:livedIn geo:England ] Order: lit:Shakespeare b:hasChild (b:c1 b:c1 b:c3).
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.