DataMining versus SemanticWeb

1. DataMining versus SemanticWeb Veljko Milutinovic, vm@etf.bg.ac.yu http://galeb.etf.bg.ac.yu/vm This material was developed with financial help of the WUSA fund of Austria.

2. DataMining versus SemanticWeb • Two different avenues leading to the same goal! • The goal: Efficient retrieval of knowledge,from large compact or distributed databases, or the Internet • What is the knowledge: Synergistic interaction of information (data)and their relationships (correlations). • The major difference: Placement of complexity

3. Essence of DataMining • Data and knowledge representedwith simple mechanisms (typically, HTML)and without metadata (data about data). • Consequently, relatively complex algorithms have to be used (complexity migratedinto the retrieval request time). • In return,low complexity at system design time!

4. Essence of SemanticWeb • Data and knowledge representedwith complex mechanisms (typically XML)and with plenty of metadata (a byte of data may be accompanied with a megabyte of metadata). • Consequently, relatively simple algorithms can be used (low complexity at the retrieval request time). • However, large metadata designand maintenance complexityat system design time.

5. Major Knowledge Retrieval Algorithms (for DataMining) • Neural Networks • Decision Trees • Rule Induction • Memory Based Reasoning, • etc… • Consequently, the stress is on algorithms!

6. Major Metadata Handling Tools (SemanticWeb) • XML • RDF • Ontology Languages • Verification (Logic +Trust) Efforts in Progress • Consequently, the stress is on tools!

7. Semantic Web Tutorial Structure (Overview) • Introduction to the Semantic Web • XML Technologies for the Semantic Web • Defining vocabularies with RDF • Ontologies and ontology languages • Challenges for the Semantic Web • References

8. World Wide Web - Today Information consumer preferences Information request preferences Search Engines (eg. Google), Information Portals Indexing, refences, collections Information and Service Providers

9. S+ S+ Request/Task Interpretation Interpretation Agents Communication, Negotiation, Planning, Decisions, Proofs Interpretation Ratings, Signatures, Certificates S+ Interpretation S+ S+ S+ Semanticly enriched information S+ „Trust“-Services Semantic Web - Vision User Preferences … Calendar … Calendar Preferences Information and Service Provider

10. A Definition of the Semantic Web “Semantic Web is an extension of the current web in which information is given well-defined meaning, better enabling computers and people to work in cooperation” Tim Berners-Lee, James Hendler, Ora Lassila, The Semantic Web, Scientific American, May 2001

11. Why? • To use the large amount of information on the Web more effectively • To enable more advanced automated processing on the Web - machines can “understand” the content • Intelligent browsers to help you find what you are looking for • To derive new information from existing information (reasoning) - Virtual global database • Advanced applications and services become possible, e.g. in - e-business - e-government - e-learning

12. Examples • Context-awareness --linking based on the meaning of the information elements • Filtering -- you could rate the pages you visit, and this is later used for automatic general recommendations • Annotations -- you could add comments to the information on the Web, and these comments can be shown to other visitors • Privatization -- you can create your own database of information from the Web

13. How? - Semantic Web layer model

14. Trusted Web Resources DAML+OIL Shared Terminology machine machine 2010 OWL XML Self Describing RDF Documents2000 HTTP Foundation of Web today1990 Human Machine HTML SGML Document Exchange Format1985 Hy Time

15. Building Blocks Semantic Web Metadata URI Data about data – labeling and structuring information in a document Universal Resource Identifier – an universal and unique name for any resource http://www.something.com/one

16. Minimalist Design • Making it as simple as possible • Simplicity helps future evolution of Semantic Web

17. Inference • Deriving new data from the existing ones • Merging data repositories gives new information • Allows the creation of more powerful applications (intelligent agents) • Unfortunately, inference can be achieved completely only when the semantics is defined formally in a language(e.g. "First Order Predicate Logic“ languages)

18. Tutorial Structure • Introduction to the Semantic Web • XML Technologies for the Semantic Web • Defining vocabularies with RDF • Ontologies and ontology languages • Challenges for the Semantic Web • References

19. XML Technologies for the Semantic Web • Overview • XML Instances • XML Document Type Definition • XML Linking • XML Schema • XML Query Language

20. What is an XML-Document ? <?xml version="1.0"?> <a> <bid="x1"> <c>David</c> <c>Marie</c> </b> <d/> <bid="x2"> <c>John</c> </b> </a> a a id=x1 id=x2 * b d b b d id c c c * c David Marie John Schema (Document Type Definition, DTD) File Format (Instance) Tree Structure Instance

21. The XML Stack Specific Applications Standardized Applications XHTML, SVG, SMIL, P3P, MathML Layout - XSL - CSS Hyperlinks - XLink - XPointer Metadata - RDF, RDFS API - DOM - SAX Schemas - XSD - Namespaces Queries - XPath - XQuery XML 1.0 Locators (URI) Unicode DTDs

22. Example of songs.xml • Example of describing a song in songs.xml using music.dtd • <song> <title>Gipsy song</title> <artist>Vlatko Stefanovski</artist> <type class=”ETHNO” /><download class=“YES”/><comments/></song> parent element defined in music.dtd child elements defined in music.dtd

23. Music.dtd Parent element • <!ELEMENT song (title, artist, album?, type, format?, download, comments?)> • <!ELEMENT title (#PCDATA)> • <!ELEMENT artist (#PCDATA)> • <!ELEMENT type EMPTY> • <!ATTLIST type • class (CLASSICAL | ROCK | POP | RAP | • JAZZ | TECHNO | ETHNO) #REQUIRED> • <!ELEMENT download EMPTY> • <!ATTLIST download • class (YES | NO) "YES" • > • <!ELEMENT comments (#PCDATA)> Child elements Attributes describe content List of values for download

24. XML Linking Simple Link Extended Link XPointer Link Group

25. XPath • A language that enables us to address parts of an XML document (elements, attributes, …) • Select the title elements of the song elements of the catalog element and all the artist elements in the document /catalog/song/title | //artist • Selects all the song elements of the catalog element that have a download element with a value of yes: /catalog/song[download=yes]/title selects any element in the document selects the child element selects several paths

26. Also… • Use * to select unknown XML elements /catalog/*/artist • Use @attribute_name to specify an attribute //song[@type=‘classical'] • XPath expressions – logical, arithmetical /catalog/song[duration<5] • XPath functions - count(), id(), last(), name(), concat(), string(), trenslate(), sum(), round(), false(), not(),… /catalog/song[last()] • To select nodes from the XML document (IE) xmlDoc.selectNodes("/catalog/song/title/text()") the path

27. XPointer • Locates portions of other XML documents (elements, attributes…), without the need to place anchors inside those documents (as in HTML) • More robust to the changes in the target document • URL + XPath • http://www.music.org/first.xml/#xpointer(//song/title[1]) URL of the document we point into XPointer expression (XPath language)

28. XML Schema • XML Schema defines a class of XML documents • Defines (explains) the datatypes, elements, and attributes • Defines and catalogues vocabularies for classes of XML documents • The document described by an XML schema can be called an instance (parallel to OOP) • The schema language, considerably extends the capabilities of XML 1.0 document type definitions (DTDs), most importantly with datatypes

29. Practically no reuseof contentmodels Syntax: Not XML Limitations of DTDs • <!ELEMENT song (title, artist, album?, type, format?, download, comments?)> • <!ELEMENT title (#PCDATA)> • <!ELEMENT artist (#PCDATA)> • <!ELEMENT type EMPTY> • <!ATTLIST type • class (CLASSICAL | ROCK | POP | RAP | JAZZ | TECHNO | ETHNO) #REQUIRED> • <!ELEMENT download EMPTY> • <!ATTLIST download • class (YES | NO) "YES"> • <!ELEMENT comments (#PCDATA)> Constructors: Elementset withContent Model Datentypes: Essentially only "String"

30. XML Schema Components • An XML Schema is comprised of a set of schema components • There are three groups of components • Primary components - Simple type definitions, Complex type definitions, Attribute declarations, Element declarations • Secondary components - Attribute group definitions, Identity-constraint definitions, Model group definitions, Notation declarations • “Helper” components – Annotations, Model groups, Particles, Wildcards, Attribute Uses

31. Example – song Type definition • <xsd:complexType name=“song" > <xsd:sequence> <xsd:element name=“title" type="xsd:string"/> <xsd:element name=“artist" type="xsd:string"/> </xsd:sequence> • <xsd:attribute name=“length" type="xsd:duration"/> </xsd:complexType> • xsd – used to denote XML Schema namespace Complex type <xsd:choice> Type declarations Simple type </xsd:choice>

32. Reusability of schemas • xs:include – to include a schema from another document (copy-paste) <xs:include schemaLocation=“collection.xsd"/> • xs:redefine – same, plus it lets you redefine schema • xs:import - reusing definitions from other namespaces (a system of libraries) <xs:import namespace="http://www.w3.org/XML/1998/namespace" schemaLocation="myxml.xsd"/> Now we can reference an external element from the imported namespace in our schema

33. Tutorial Structure • Introduction to the Semantic Web • XML Technologies for the Semantic Web • Defining vocabularies with RDF • Ontologies and ontology languages • Challenges for the Semantic Web • References

34. Defining vocabularies with RDF • Motivation for RDF • RDF Instances • Basic concepts and building blocks • Syntax options • Reification • Collections • RDF Schema: Defining your own Vocabularies • Supporting Interoperability with RDF

35. What do we NOT get from XML? • Superimposing (meta) information: • XML combines metainformation and content • Datatypes that we can „reason“ about: • Example:CLASSICAL | ROCK | POP | RAP | JAZZ | TECHNO | ETHNOis just a choice of allowed strings. We cannot represent that DIXIE is a subclass of JAZZ, BLUES overlaps with ROCK, ETHNO • Bottom up reuse of vocabularies • Independently evolved XML Schemas for one and the same thing • How do you model an „address“?

36. RDF: Defining Semantics on the Web • There is a need to describe resources on the Web in a form that can be interpreted by machines across the Web • Interpretation depends on the context of a resource eg. Jaguar (car vs. beast) • Using their experience and cognitive abilities humans may infer the context of a resource in many ways, even if it is not made explicit • Software can interpret context only if it is described explicitly and formally • RDF and the ontology languages building upon RDF provide means to explicate (part of) this context

37. RDF-Resource Description Framework • Defines a framework for structuring and describing resources like documents in the Semantic Web • Enables the definition of vocabularies for the description of resources in an application domain; • Goals: • Extensibility, interoperability, and reuse of vocabularies; • Improved support for interpretation of data by machines

38. The RDF Data Model • Simple but powerful datamodel for the description of resources and the creation of metadata • Consists of three core concepts: • Resource • Property • Statement + Class(in RDF Schema) • Similar to other modeling approaches (e.g. object-oriented modeling), but property-centric, not class-centric

39. RDF Statement and Graph • Each triple (S, P, O) node - arc - node represents an RDF statement Gipsy song is performed by Vlatko Stefanovski. subject (resource) object (resource or literal) predicate (property) http://www.music.org/songs/g/gipsySong http://www.artist.org/stefanovski Performed by Song represented by entry in a (fictive) song directory Artist represented by his homepage

40. Arcs in the RDF Graph An Arc • represents the predicate of an RDF statement • is labeled with a URI referring to an RDF property • is directed pointing from the subject of a statement to the object of a statement object subject predicate http://www.artist.org/stefanovski http://www.music.org/songs/g/gipsySong music:performed by

41. RDF Resource • The Resource forms the central concept in RDF • Anything that can be described can act as a resource • Web page, part of web page, web site, book, photograph, persons, … • Resources are identified by a resource identifier - URI (plus optional anchor IDs) • Compare for an entity (in the Entity Relationship model) or an object (in an object-oriented model)

42. RDF Property An RDF Property is used to express • A characteristic of an resource or • A binary relation between resources • A predicate in a statement • A property can be compared to a (binary) relationship among entities (in the Entity Relationship model)

43. Example • The individual whose name is Vlatko Stefanovski and whose email is V.Stefanovski@artists.org, is the artist of http://www.music.org/songs/g/gipsySong URI reference http://www.music.org/songs/g/gipsySong blank node music:artist node person:name person:homepage Vlatko Stefanovski http://www.artists.org/stefanovski literal

44. XML Serialization • How to translate the RDF graph structure into XML’s tree-oriented notation <rdf:Description rdf:about = “http://www.music.org/songs/g/gipsySong”> <music:performedby> <rdf:Description> <person:name>Vlatko Stefanovski</person:name> <person:homepage> <rdf:Description about = “http://www.artists.org/stefanovski”> </rdf:Description> </person:homepage> </rdf:Description> </music:performadby> </rdf:Description> http://www.music.org/songs/g/gipsySong music:performedby person:name person:homepage http://www.artists.org/stefanovski Vlatko Stefanovski

45. Reification • Latin: Res ... Thing -> Reification ... “Thing Making“ • Statements themselves can be considered as resources (things) in RDF. Thus, it is possible to make statements about statements (Reification). • Possible applications: • Definition of a context for a statement with respect to time, place, validity, …. • Embed a statements into a discourse (claims, doubts, proofs of statements) • … Example: Statement A: <sonata XY> <composer> <Mozart> Statement B: <music expert A> <claims> <statement A> <music expert C> <doubts> <statement A>

46. Reification Syntax • The statement to be reified has to be modeled as an RDF resource; • The RDF vocabulary provides special constructs for this purpose: • The class rdf:Statements which is the type of all RDF statements. • The property rdf:type which is used to associate an RDF resource with a class. • The property rdf:subject refers to the subject of the modeled statement (i.e. to the described resource) • The property rdf:predicate refers to the property used as a predicate in the modeled statement • The property rdf:object refers to the object of the modeled statement (i.e. the property value)

47. How to create a reified statement? • Associate the subject, predicate and object of the statement with the resource rdf: Statement This is done by using the rdf:subject, rdf:predicate and rdf:object properties; rdf:Statement rdf:type rdf:subject rdf:predicate rdf:object music:composer www.operas.org/Zauberflöte www.operas.org/Zauberflöte www.artists.org/Mozart www.artists.org/Mozart music:composer

48. How to create a reified statement? • Now the created node which represents the statements can be used as an object or subject of another RDF statement Statement becomes a resource www.musicExperts.org/ExpertA rdf:Statement music:claimedBy rdf:type rdf:subject rdf:predicate rdf:object music:composer www.operas.org/SonataXY www.artists.org/Mozart

49. XML Syntax for Reification <rdf:RDF xmlns:rdf = "http://w3.org/1999/02/22-rdf-syntax-ns#" xmlns:music="http://ipsi.fhg.de/music-schema#"> • <rdf:Description> <rdf:type resource=" http://w3.org/1999/02/22-rdf-syntax-ns#Statement” > <rdf:subject resource="http://www.operas.org/SonataXY " /> <rdf:predicate resource="http://ipsi.fhg.de/music-schema#Composer" /> <rdf:object resource = “http://www.artists.org/Mozart” /> <music:claimedBy resource = “http://www.musicExperts.org/ExpertA” /> </rdf:Description> </rdf:RDF> Property of the statement

50. RDF Collections • An RDF Container models a collection of resources. • The RDF model supports three types of containers: • Bag - an unordered list of resources or literals. • Sequence - An ordered list of resources or literals. • Alternative - A list of resources or literals that represent alternatives for the (single) value of a property. • Bag and Sequence can be used for multivalued properties