1. Semantic Web for the Working Ontologist Dean Allemang Jim Hendler SNU IDB laboratory

2. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Working Ontology Contents 2 ■ Chapter 1 What is the Semantic Web? ■ Chapter 2 Semantic Modeling ■ Chapter 3 RDF-The Basis of the semantic Web ■ Chapter 4 Semantic Web Application Architecture ■ Chapter 5 RDF and Inferencing ■ Chapter 6 RDF Schema ■ Chapter 7 RDFS-Plus ■ Chapter 8 Using RDFS-Plus in the Wild ■ Chapter 9 Basic OWL ■ Chapter 10 Counting and Sets in OWL ■ Chapter 11 Using OWL in the Wild ■ Chapter 12 Good and Bad Modeling Practices ■ Chapter 13 OWL Levels and Logic

3. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Chapter 5 RDF and Inferencing ■ Inference in the Semantic Web Virtues of Inference-Based Semantics ■ Where are the Smarts? Asserted Triples versus Inferred Triples When Does Inferring Happen? Inferencing as Glue 3

4. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Inference in the Semantic Web 4 chamois search shirts category results nothing chamois search Henleys category results

5. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Inference in the Semantic Web ■ One solution to this problem is to leverage the power of the query E.g. “Show me all items in category ‘Shirts’, or in any subcategory of ‘Shirts’, or any sub-subcategory of ‘Shirts’, and so on” ■ Semantic Web provides a model of data expression for representing the relationships between data items In this sense, it allows a data modeler to create data that are more connected, better integrated, and smarter 5

6. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Inference in the Semantic Web ■ To make data seem more connected and consistently integrated, we must be able to add relationships into data 6 “shirts” is a broader term than “Henleys” “Henleys” is a subclass of the “Shirts”

7. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Inference in the Semantic Web ■ Semantic Web infrastrucutre provides a formal and elegant specification of the meaning of the various terms like subClassOf “B is subClassOf C” ▶ “Every member of class B is also a member of class C” ▶ This specification is based on the notion of inference − “x is a member B” → “x is a member of C” ■ The pattern for the subClassOf in RDFS 7 IF ? A rdfs:subClassOf ?B AND ?x rdf: type ?A THEN ?x rdf: type ?B type propagation

8. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Inference in the Semantic Web ■ Virtues of Inference-Based Semantics A major concern for making our data more useful is to have they behave in a consistent way Taking subClassOf as an example, if a single class to be specified as subClassOf two other classes ▶ In an informal thesaurus setting, it needs discussions and design decisions ▶ In an inference-based system, the answer to the question is defined by the interaction of the basic inference patterns − If A is subClassOf B and A is also subClassOf C, then any individual x that is a member of A will be a member of B and C 8

9. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Inference in the Semantic Web ■ A new component is added to the deployment system to make the application behave differently The component will respond to queries based not only on the triples but also on the triples that can be inferred based on the rules of inference New item: inferencing component 9 Application RDF Files Analytics Interface … Converters and Scrapers Parser and Serializer RDF Store (merge) Query Engine Webpages, Spreadsheets, Tables, Databases, etc.

10. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Where are the Smarts? ■ Asserted Triples versus Inferred Triples Asserted triples, as the name suggested, are the triples that were asserted in the original RDF store Inferred triples are the additional triples that are inferred by one of the inference rules 10 Tip ： if the inference engine infers a triple that has already been asserted, we will consider the triple to have been asserted

11. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Where are the Smarts? ■ Example 11 shop:MensWear shop:Oxfords shop:Shirts shop:Henleys shop:Tshirts shop:ChamoisHenley shop:ClassicOxford rdfs:subClassOf rdf:type

12. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Where are the Smarts? ■ Example shop:Henleys rdfs:subClassOf shop:Shirts. shop:Shirts rdfs:subClassOf shop:MensWear. shop:Blouses rdfs:subClassOf shop:WomensWear. shop:Oxfords rdfs:subClassOf shop:Shirts. shop:Tshirts rdfs:subClassOf shop:Shirts. shop:ChamoisHenley rdfs:type shop:Henleys. shop:ClassicOxford rdfs:type shop:Oxfords. shop:ClassicOxford rdfs:type shop:Shirts. shop:BikerT rdfs:type shop:TShirts. shop:BikerT rdfs:type shop:Menswear. shop:ChamoisHenley rdf:type shop:Shirts. shop:ChamoisHenley rdf:type shop:MensWear. shop:ClassicOxford rdf:type shop:Shirts. shop:ClassicOxford rdf:type shop:MensWear. shop:BikerT rdf:type shop:Shirts. shop:BikerT rdf:type shop:MensWear. 12 inference

13. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Where are the Smarts? ■ Example 13 shop:MensWear shop:Oxfords shop:Shirts shop:Henleys shop:Tshirts shop:ChamoisHenley shop:ClassicOxford Inferred triples

14. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Where are the Smarts? ■ When does Inferencing happen? The simplest approach ▶ Quite simple to describe and implement ▶ Risks an explosion of triples in the triple store 14 Store all triples in a single store, regardless of whether they are asserted or inferred As soon as pattern is identified, any inferred triples are inserted into the store Asserted & Inferred inferencing

15. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Where are the Smarts? ■ When does inferencing happen? Another approach ▶ Risks duplicating inference work ▶ Parsimonious in terms of persistent storage 15 Never store any inferred triples in any Persistent store at all Inferencing is done in response to queries only Asserted inferencing query

16. Semantic Web for the Working OntologistDean Allemang, Jim Hendler Where are the Smarts? ■ Inferencing is the glue that holds the Semantic Web together ■ Two fundamental component in data integration A model that expresses the relationship between the two data sources ▶ Model consists of a single class that has both of the classes to be integrated as subclasses → subClassOf Notion of inferencing ▶ The process of inferencing that applies the model to the two data sources to produce a single, integrated answer ■ Inference systems are very useful for making data more consistent and connected 16