Semantic Web Introduction

1
Semantic WebIntroduction
2
Outline

• The current Web
• What is Semantic Web?
• How the Semantic Web Will Be Possible?
• Framework of the Semantic Web

3
Weaving the Web Vision of Tim Berners-Lee

• The first step is putting data on the Web in a
  form that machines can naturally understand, or
  converting it to that form. This creates what I
  call a Semantic Web (SW) a web of data that can
  be processed directly or indirectly by
  machines-- Tim Berners-Lee Weaving the Web,
  Harper San Francisco, 1999
• Require that there be a machine-understandable
  semantics for some or all of the information
  presented in the WWW

4
Web Today

• Information repository.
• Simplicity.
• Primarily for human interpretation and use.

5
Current Web for Knowledge Management

• Search information Keyword-based search
• Extracting information human browsing and
  reading
• Automatic agents lack the commonsense knowledge
  required to extract "relevant" from textual
  representations and fail to integrate information
  spread over different source
• Maintenance
• keeping collections consistent, correct, and up
  to date requires a mechanized representation of
  semantics and constraints that help to detect
  anomalies
• Automatic document generation
• According to user profiles or other relevant
  aspects
• Require a machine-accessible representation of
  the semantics of these information sources

6
Current Web for Knowledge Management (Cont.)

• Semantic Web technology will enable structural
  and semantic definitions of documents providing
  completely new possibilities
• Intelligent search instead of keyword matching
• Query answering instead of information retrieval
• Document exchange among departments via ontology
  mappings
• Definition of customized views on documents

7
Current Web for Web Commerce

• B2C online store, auction houses, online
  marketplaces
• Shop-bots visit several stores, extract product
  information, and present to the customer an
  instant market overview
• Functionality is provided via wrappers written
  for each online store
• Keyword search regularities in the presentation
  format of stores' Web sites text extraction
  heuristics
• Efforts time-consuming activity for writing
  wrappers
• Quality limited product information, error
  prone, incomplete
• Why most product information on Web sites is
  provided in natural language, and automatic text
  recognition is still a research area with
  significant unsolved problems
• Require a machine-processable semantics for the
  information provided

8
Current Web for Web Commerce (Cont.)

• When standard representation formalisms for the
  structure and semantics of data are available
• Software agents can then be built that can
  understand the product information the Web sites
  provide
• Meta-online stores can then be constructed with
  little effort, and this technique will also
  enable complete market transparency in various
  dimensions of diverse product properties
• The low-level programming of wrappers based on
  text extraction and format heuristics will be
  replaced by semantic mappings that translate
  different formats used to represent products and
  can be used to navigate and search automatically
  for the required information

9
Toward Semantic Web
Information repository ? Information-service
provider
10
What Is the Semantic Web?
11
Vision of Tim Berners-Lee for the Future of the
Web

• Two-part vision
• Make the Web a more collaborative medium
• Make the Web understandable, and thus
  processable, by machines
• Original Web proposal to CERN
• Relations between information items like
  "includes," "describes,", and "wrote" ? not
  currently captured on the Web
• Use RDF to capture such relationship

12
Original Web Proposal to CERN
Encompass additional metadata above and beyond
what is currently in the Web. This additional
metadata is needed for machines to be able to
process information on the Web
13
Web vs. Semantic Web
Web
Semantic Web
14
Smart Data

• How do we create a web of data that machines can
  process?
• MAKE THE DATA SMARTER
• The smart data continuum
• Text and databases (pre-XML)
• XML documents for a single domain
• Taxonomies and documents with mixed vocabularies
• Ontologies and rules

15
The Smart Data Continuum
16
Text and databases

• Pre-XML
• The initial stage where most data is proprietary
  to an application
• The "smarts" are in the application and not in
  the data

17
XML documents for a single domain

• The stage where data achieves application
  independence within a specific domain
• Data is now smart enough to move between
  applications in a single domain
• Example XML standards in the healthcare
  industry, insurance industry, or real estate
  industry

18
Taxonomies and documents with mixed vocabularies

• Data can be composed from multiple domains and
  accurately classified in a hierarchical taxonomy
• The classification can be used for discovery of
  data
• Simple relationships between categories in the
  taxonomy can be used to relate and thus combine
  data
• Data is now smart enough to be easily discovered
  and sensibly combined with other data

19
Ontologies and rules

• New data can be inferred from existing data by
  following logical rules
• Data is now smart enough to be described with
  concrete relationships, and sophisticated
  formalisms where logical calculations can be made
  on this "semantic algebra."
• This allows the combination and recombination of
  data at a more atomic level and very fine-grained
  analysis of data
• Data no loner exists as a blob but as a part of a
  sophisticated microcosm
• Example automatic translation of a document in
  one domain to the equivalent (or as close as
  possible) document in another domain

20
Definition of Semantic Web

• A machine processable web of smart data
• Smart data can be further defined as data that is
  application-independent, composeable, classified,
  and part of a larger information ecosystem
  (ontology)
• An extension of the current web in which
  information is given well-defined meaning, better
  enabling computers and people to work in
  cooperation.
• An infrastructure enables machines to COMPREHEND
  semantic documents and data.
• A brain for humankind, which assists the
  evolution of human knowledge as a whole.

21
How the Semantic Web Will Be Possible?
22
Achieving a Semantic Web Requires

• Developing languages for expressing
  machine-understandable meta-information for
  documents and developing terminologies (i.e.
  namespaces or ontologies) using these languages
  and making them available on the Web
• Developing tools and new architectures that use
  such languages and terminologies to provide
  support in finding, accessing, presenting, and
  maintaining information sources
• Realizing applications that provide a new level
  of service to the human users of the semantic Web

23
Languages Two Aspects

• Languages that provide formal syntax and formal
  semantics to enable automated processing of
  content
• Languages that provide standardized vocabulary
  referring to real-world semantics enabling
  automatic and human agents to share information
  and knowledge ontologies

24
Formal Languages

• Layer language model for the Semantic Web
• HTML just for presentation, not for processing
• XML separate content and data from presentation
• The Semantic Web is an XML application
• RDF defines a syntactic convention and a simple
  data model for representing machine-processable
  semantic of data
• RDF Schema (RDFS) defines basic ontological
  modeling primitives on top of RDF
• Ontology Inference Layer (OIL) and DARPA Agent
  Markup LanguageOntology (DAML-ONT) full blown
  ontology modeling language as extension of RDFS

25
Data representation

• XML addresses only document structure.
• RDF is a web-enabled language of Subject, Verb,
  Object triples to represent relationships between
  data.
• Ontology is the specification of a
  conceptualizationdefines terms and relationships
  between them.
• Logic allows us to reason across the RDF elements.

26
How does XML Fit into the Semantic Web?

• XML is the syntactic foundation layer of the
  Semantic Web
• XML guarantees a base level of interoperability
• XML is built upon Unicode characters and Uniform
  Resource Identifiers (URIs)
• Unicode allow XML to be authored using
  international characters
• URI used as unique identifiers for concepts in
  Semantic Web
• XML is not enough (only syntactic
  interoperability)
• Sharing XML documents adds meaning to the
  content but, only when both parties understand
  the element name
• ? ltpricegt12.00lt/pricegt VS. ltcostgt12.00lt/costgt
• ? Require SW technologies like ontologies

27
RDF and RDFS

• RDF standard for Web metadata developed by W3C
• Suitable for describing any Web resources
• Provide interoperability among applications that
  exchange machine-understandable information on
  the Web
• An XML application and adds a simple data model
  on top of XML
• Objects, properties, and values of properties
• RDFS candidate recommendation
• Define additional modeling primitives on top of
  RDF
• Allow the definition of classes (i.e. concepts),
  inheritance hierarchies for classes and
  properties, and domain and range restrictions for
  properties

28
RDF Application Integration Hub
29
Logical Assertions

• An assertion is the smallest expression of useful
  information
• Use Resource Description Framework (RDF) to
  capture the association (assertion) between
  subjects and objects
• How can we use these assertions?
• The author of a document has written other
  articles on similar topics
• A well-know authority on the subject has refuted
  the main points of an article
• Assertions are not free-form commentary but
  instead add logical statements to a resource or
  about a resource

Object
OnlineMike
knows
Subject
OnlineMary
age
Each Subject/ Object is a resource
Literal
33
30
Classification Taxonomy

• We classify things to establish groupings by
  which generalization can be made
• Downside of classification systems
• Categories can be arbitrary
• Membership criteria are often ambiguous
• Lack rigorous logic for machines to make
  inference from useful for humans browsing for
  information
• XML Topic Maps (XTM)

• Example Linnaean classification of a house cat
• Kingdom Animalia
• Phylum Chordata
• Class Mammalia
• Order Carnivora
• Family Felidae
• Genus Felis
• Species Felis domesticus

31
Ontologies Formal Class Model

• An ontology is a formal, explicit, specification
  of a shared conceptualization
• Conceptualization an abstract model of some
  phenomenon in the world that identifies the
  relevant concepts of that phenomenon
• Explicit the type of concepts used and the
  constraints on their use are explicitly defined
• Formal ontology should be machine
  understandable
• Shared an ontology captures consensual
  knowledge. It is not restricted to some
  individual but accepted by a group
• Ontology examples
• WordNet (http//www.cogsci.princeton.edu/wn)
  a thesaurus for over 100,000 terms explained in
  natural language
• CYC (http//www.cyc.com) formal axiomating
  theories for many aspects of common sense
  knowledge

32
Ontologies (Cont.)

• An ontology is a formal representation of classes
  and relationships between classes to enable
  inference
• Formal class hierarchies constrained properties
  relations between classes
• An ontology contain classes, subclasses,
  properties of classes, and relations between
  classes
• An ontology captures logical information in a
  manner that can allow inference
• John is a Leader ? John is a Person and John may
  lead an organization
• Additional formalisms are added to enable
  inference
• Symmetric property, transitive property
  (hasAncestor)

33
Key Ontology Components
Person Birthdate date Gender char
Image
depiction
knows
published
is-A
worksfor
Resource
Organization
Leader
leads
34
Ontologies (Cont.)

• Shared formal conceptualizations of particular
  domains.
• Enable Web-based knowledge processing, sharing,
  and reuse.
• Provide a common understanding of topics.
• Ensure that everyone agrees on terms, types,
  constraints, etc.

35
Ontologies (Cont.)

• Developed in AI to facilitate knowledge sharing
  and reuse
• Applications knowledge engineering, NLP,
  knowledge representation, intelligent information
  integration, cooperative information systems,
  information retrieval, electronic commerce, and
  KM
• Why so popular what they promise a shared and
  common understanding of some domain that can be
  communicated among people and application systems
• Aim at consensual domain knowledge cooperative
  development process

36
Rules

• With XML, RDF, and inference rules, the Web can
  be transformed from a collection of documents
  into a knowledge base
• Modus Ponens "If P is TRUE, then Q is TRUE"
• P is TRUE therefore, Q is TRUE
• "An apple is tasty if it is not cooked. This
  apple is not cooked. Therefore, it is tasty."
• SW can use information in an ontology with logic
  rules to infer new information
• "Let's say one company decides that if someone
  sells more than 100 of our products, then they
  are a member of the Super Salesman Club." ? a
  smart program can now follow this rule to make a
  simple deduction "John has sold 102 things,
  therefore John is a member of the Super Salesman
  club"

37
Using rules to infer the uncleOf Relation

• If a person C is a male and childOf a person A,
  then person C is a "sonOf" person A
• If a person B is a male and siblingOf a person A,
  then person B is a "brotherOf" person A
• If a person C is a "sonOf" person A, and person B
  is a "brotherOf" person A, then person B is the
  "uncleOf" person C

PersonA
PersonB
siblingOf
childOf
uncleOf
PersonC
38
OIL An Ontology Language

• OIL (http//www.ontoknowledge.org/oil)
• Ontology Inference Layer or Ontology Interchange
  Language
• Three requirements for an ontology language
• It must be highly intuitive to the human user.
• Given the current success of the frame-based and
  OO modeling paradigm, it should have a frame-like
  look and feel
• It must have a well-defined formal semantics with
  established reasoning properties in terms of
  completeness, correctness, and efficiency
• It must have a proper link with existing Web
  languages like XML and RDF, ensuring
  interoperability
• OIL fulfill the above three requirements

39
DAML-ONT Another Ontology Language

• DAML-ONT (http//www.daml.org)
• Funded by the U.S. DARPA
• Sill in an early stage of development and lacks a
  formal definition of its semantics

40
Tools

• Formal languages to express and represent
  ontologies
• Editors and semiautomatic construction to build
  new ontologies
• Reusing and merging existing ontologies (ontology
  environment)
• Reasoning services (instance and schema
  inferences that enable advanced query answering
  service, support ontology creation, and help map
  between different terminologies)
• Annotation tools to link unstructured and
  semi-structured information sources with metadata
• Tools for information access and navigation that
  enable intelligent information access for human
  users
• Translation and integration services between
  different ontologies that enable multi-standard
  data interchange and multiple view definitions
  (especially for B2B electronic commerce)

41
Ontology Editors

• Help human knowledge engineers build ontologies
• Ontology development and maintenance
• Define concept hierarchies, attributes for
  concepts, axioms and constraints
• Inspect, browse, codify, and modify ontologies
• GUI
• Conform to existing standards in Web-based
  software development
• Example Protégé (Stanford)

42
Protégé Editor
43
Semi-Automatic Ontology Constructor

• Manually building ontologies is time-consuming
• Tools that learn ontologies from natural language
  exploit the interaction constraints on the
  various levels (morphology, syntactic, semantic,
  pragmatics, background knowledge) in order to
  discover new concepts and stipulate relationships
  among concepts
• These tools combine machine learning, information
  extraction, and linguistic techniques to extract
  relevant concepts, build is-a hierarchies, and
  determine relationships among concepts

44
Semi-Automatic Ontology Constructor Text-To-Onto

• KM group of the Institute AIFB, Karlsruhe
  University
• Learn ontologies from text
• Select a relevant corpus of domain texts (NLP
  texts or HTML texts)
• Use domain lexicon to perform domain-specific
  parsing
• Existing knowledge structures (ex. A taxonomy of
  concepts) are incorporated as background
  knowledge
• Discover new knowledge structures, which are then
  captured in the ontology modeling module to
  expand the existing ontology

45
Text-To-Onto
46
Ontology Environment

• Reuse existing ontologies to save time and labor
• Must allow adaptation and merging of existing
  ontologies to make them fit for new tasks and
  domains
• Operations for combining ontologies ontology
  inclusion, restriction, and polymorphic
  refinement
• Use ontologies in different formats, reorganize
  taxonomies, resolve name conflicts, browse
  ontologies, edit terms
• Example Chimaera (Stanford) merging and
  diagnosing (and evolving) ontologies

47
Chimaera
48
Reasoning Services

• Reasoning over instances of an ontology
• Derive a certain value for an attribute applied
  to an object
• Powerful support in formulating rules and
  constraints and answering queries over schema
  information
• Used to answer queries about the explicit and
  implicit knowledge specified by an ontology
• Help to build ontologies
• Example Ontobroker (Ontoprise,
  http//www.ontoprise.de)

49
Reasoning Services (Cont.)

• Reasoning over concepts of an ontology
• Automatically derives the right position for a
  new concept in a given concept hierarchy
• Help to build ontologies
• Example FaCT (Fast Classification of
  Teminologies) (Manchester University) derive
  concept hierarchies automatically

50
Annotation Tools

• Ontologies can be used to describe a large number
  of instances
• Annotation tools help the knowledge engineer to
  establish such links via
• Linking an ontology with a database schema or
  deriving a database schema from an ontology (in
  cases of structured data)
• Deriving an XML DTD, an XML schema, and an RDF
  schema from an ontology (in case of
  semi-structured data)
• Manually or semi-automatically adding ontological
  annotation to unstructured data

51
The Ontology-Learning Process
52
Tools for Information Access and Navigation

• Low-level navigation now clicking on links and
  using keyword searches
• Keyword-based search retrieves irrelevant
  information that uses a particular word in a
  different meaning from the one intended, and it
  may miss relevant links in which different words
  than the keyword are used to describe the content
  for which the user is searching
• Navigation is supported only by predefined links
  current navigation technology does not support
  clustering and linking of pages based on semantic
  similarity
• Query responses require human browsing and
  reading to extract the relevant information from
  the information sources returned
• Burden Web users with an additional loss of time
  and seriously limits information retrieval by
  automatic agents

53
Tools for Information Access and Navigation
(Cont.)

• Low-level navigation now (Cont.)
• Keyword-based document retrieval fails to
  integrate information spread over different
  sources
• Current retrieval services can retrieve only
  information that is directly represented on the
  Web. No further inference service is provided for
  deriving implicit information that must be
  derived from the explicit text
• Ontologies can
• Support IR based on the actual content of a page
• Help user navigate information space based on
  semantic concepts
• Enable advanced query answering and information
  extraction services, integrating heterogeneous
  and distributed information sources enriched by
  inferred background knowledge

54
Hyperbolic Browsing Interface
Semantic Information Visualization
55
Automatically Generated Semantic Structure Map
56
Translation and Integration Services

• Current B2B the heterogeneity of product
  descriptions on Web sites and the exponentially
  increasing effort that must be devoted to mapping
  these heterogeneous descriptions as the number of
  Web sites increases
• Effective and efficient CM of heterogeneous
  product catalogues is critical for the success of
  B2B
• Mapping heterogeneous descriptions
• Different representations of product catalogs
  must be merged
• XML with DTD1 ?? XML with DTD2
• Different vocabularies used to describe products
  must be merged
• Languages, concepts, attributes, values and value
  types

57
Applications

• Applications for Knowledge Management
• On-To-Knowledge
• SwissLife (http//www.swisslife.ch)
• British Telcom (http//www.bt.com/innovations)
• Enersearch (http//www.enersearch.se)
• Applications for B2C
• Semantic Edge (http//www.semanticedge.com)

58
Web Services

• Web services are software services identified by
  a URI that are described, discovered, and
  accessed using Web protocol
• Web services consume and produce XML
• Furthering the adoption of XML, or more smart
  data
• As Web services proliferate, they are more
  difficult to discover ? using SW technologies to
  solve the Web service discovery problem
• Enabling Web services to interact with other Web
  services

59
Semantic Web services and agents

• Automation tasks
• Service discovery
• Service comparison
• Service execution
• Service composition and interoperation
• What should be markup?
• Web services
• User constraints and preferences
• Agent procedure

60
Framework of Semantic Web
61
Semantic Web Principles

• Everything is identifiable
• Any abstract thing can have a URI.
• Partial information
• Anyone can say anything about anything.
• Dont expect global consistency.
• Evolution
• Allow effective combination of the independent
  work of diverse communities.
• Minimalist design
• Make the simple things simple, and the complex
  things possible.

62
What achieves the Semantic Web?

• Knowledge representation
• Ontology Logic Inference Learning
• Semantic Web services
• Agents

63
Semantic interoperability
64
Proofs exchange between agents
65
Trust

• Add semantics to make trust determination better
• You may want to allow access to information if a
  trusted friend vouches (via a digital signature)
  for a third party
• Digital signatures are crucial to the "web of
  trust"
• By allowing anyone to make logical statements
  about resources, smart applications will only
  want to make inferences on statements that they
  can trust
• Verifying the source of statements is a key part
  of SW

66
Semantic Web wave
67
Who are involved?

• Artificial intelligence
• The web
• Databases
• Agents
• Theoretical computer science and logic
• Systems
• Computational linguistics and pattern recognition
• Document engineering and digital libraries
• Human-computer interfaces
• Social and human sciences

68
Discussions

• The Semantic Web is not a separate Web but an
  extension of the current one.
• Markup makes data computer-interpretable,
  use-apparent, and agent-ready.
• The Semantic Web will break out of the virtual
  realm and extend into our physical world.
• Ubiquitous knowledge and computing
• Can you imagine the Semantic Web A brain for
  humankind?

69
References

• http//www.w3.org/2001/sw/
• http//www.semanticweb.org/
• Semantic Web-enabled Web services
  (http//swws.semanticweb.org)