Semantic Web
PRESENTED BY: ANDRÉ MAZAYEV
ANDRE.MAZ.90[ AT]GMAIL[DOT]COM
Semantic Web? Whaaat?
◦ What semantic web means?
◦ Smarter web!! Duuuh!
Semantic Web? Whaaat?
◦ What semantic web means?
◦ Smarter web!! Duuuh!
◦ Ok. But more specifically?
◦ It’s a web where it is easier to find stuff on internet
Semantic Web? Whaaat?
◦ What semantic web means?
◦ Smarter web!! Duuuh!
◦ Ok. But more specifically?
◦ It’s a web where it is easier to find stuff on internet
◦ Yeah! But how?
◦ Hmmmmm……
Web 2.0
◦ Search Process
◦ Refine search as you go
◦ The user is guiding the search accordingly to the results that are shown
◦ Search engine is only performing syntax based pattern match
◦ Plus some features to improve performance and accuracy
◦ Semantics are not used or used in a limited way during the search process
Syntax and Semantics
◦ Syntax
◦ About form
◦ Semantics
◦ About meaning
Syntax and Semantics
◦ Syntax
◦ Green, Yellow, Red
◦ Semantics
◦ Green = Go
◦ Yellow = Better stop
◦ Red = Stop
Traffic Light
Adapted from: Semantic Web from the 2013 Perspective
User’s Web Example
Example of dumb web
◦ Goal
◦ Find the telephone number of James Bond
User’s Web Example
Example of dumb web
◦ Goal
◦ Find the telephone number of James Bond
◦ For humans the answer is easy to find
◦ James Bond’s telephone number is 1-800-555-0199
◦ James Bond is a fictional MI6 agent
◦ Since it’s a fictional agent we can infer that the number must be fake
Machine’s Web Example
Example of dumb web
Source code of dumb web
◦ For machines find Bond’s number is a hard task
◦ No machine “readable” semantics
◦ Current Web
◦ Created for document sharing
◦ Instead of data sharing
◦ Adapted for Human to Human
◦ Machine to Machine communication is difficult
Smart vs Dumb Web
Example of dumb web
Example of smart web
Smart vs Dumb Web
Visually both pages are identical
Smart page carries much more
“meaning”
Example of dumb web
Example of smart web
Smart vs Dumb Web
Source code of smart webSource code of dumb web
Source code analysis
Contains more machine friendly structure
◦ Vocabulary is defined
◦ Data is structured
◦ Data is enriched
The data can be represented as a graph
Source code of smart web
James
Bond
1-800-555-0199
James Bond
typeof
name
telephone
Person
Source code analysis
Source code of smart web
James
Bond
1-800-555-0199
James Bond
typeof
name
telephone
Person
Source code analysis
With structured data it’s easy for a machine to find Bond’s telephone number
Source code of smart web
Graph analysis
◦ Simple statements
◦ Subject – Predicate – Object
◦ All elements have their own URL
◦ Data is structured
◦ Data can be explored by machines
James
Bond
1-800-555-0199
James Bond
typeof
name
telephone
Person
URL
URL
URL
Structured Data Tool
Source code of a page with semantic markup Extracted data
Structured Data Tool
Extracted data
◦ Data recognized by Google’s web crawler
◦ With structured data answers are easy to get
◦ What?
◦ Where?
◦ When Open?
Semantic Web
Present Future
Web of Documents Web of Data
Small Change
Big Difference
◦ Data is explicit
◦ Data is connected
◦ Data can be explored by machines
◦ Nontrivial connections can be found
◦ Demo
◦ RelFinder
Semantic Building Blocks
RDF
RDF
◦ Resource Description Framework
◦ Simple statements (triples)
◦ Subject – Predicate – Object
◦ Building block of RDFS and OWL
◦ Multiple serialization formats
◦ RDF/XML
◦ Turtle
◦ N-Triples
Bond example in Turtle
RDFS
RDFS
◦ RDF Schema
◦ Limited expressivity
◦ Describes classes, subclasses and properties
◦ Primary focused on “is a” and “sub class of”
relationships
Vocabulary
Canine
Animal Human
Mammal
Feline
Reptile
Taxonomy
Animal
MammalReptile
Human
Canine Feline
subClassOf subClassOf
subClassOf subClassOf
subClassOf
SPARQL
SPARQL
◦ SPARQL Protocol And RDF Query Language
◦ SQL-Like structure
James
Bond
1-800-555-0199
James Bond
typeof
name
telephone
Person
Graph
SPARQL
◦ SPARQL Protocol And RDF Query Language
◦ SQL-Like structure
James
Bond
1-800-555-0199
James Bond
typeof
name
telephone
Person
Graph
Goal: Find Bond’s Number
SPARQL
◦ SPARQL Protocol And RDF Query Language
◦ SQL-Like structure
James
Bond
1-800-555-0199
James Bond
typeof
name
telephone
Person
Graph
Query
Goal: Find Bond’s Number
SPARQL
◦ SPARQL Protocol And RDF Query Language
◦ SQL-Like structure
James
Bond
1-800-555-0199
James Bond
typeof
name
telephone
Person
Graph
Answer
Query
Goal: Find Bond’s Number
OWL
OWL
◦ Web Ontology Language
◦ Highly expressive
◦ Brings expressivity of logic to Semantic Web
◦ More expressive than RDFS
◦ Allows to express
◦ Constraints
◦ Cardinality
◦ Unions
◦ Intersections
◦ Etc.
Resource that has property hasParent with value
Bond belongs to a class named BondChild
OWL Restriction
Note: Often the concepts of taxonomies and ontologies overlap and used to describe same thing
SWRL
SWRL
◦ Semantic Web Rule Language
◦ Combines parts from OWL and Datalog
◦ Rule syntax
◦ If body (antecedent) then assert head (consequent)
x3 is x1’s uncle
Under Development
◦ Pending questions
◦ How to ensure security of data?
◦ How to validate new data?
◦ Is source data reliable?
Data Silos
◦ Each application has its own
◦ Goals
◦ Vocabularies
◦ Knowledge base
◦ Not integrated with other data systems
◦ May have overlapping data
Application 1
Application 2
Application 3
Sensor
Network
Gateway
Server Application
Data
Source
Relational
DB
Relational
DB
Semantic Bridges
Sensor
Network
Gateway
Server
Data
Source
Relational
DB
Relational
DB
RDB
Parser
CSV
Parser
WEB
Parser
RDF
Interfaces
Combined RDF
Model
Combined
Knowledge Model
Application 1
Application 2
Application 3
Application
RDB
Parser
CSV
Parser
WEB
Parser
Model Knowledge Model
Application 1
Application 2
Application 3
er
RDB
Parser
CSV
Parser
WEB
Parser
RDF
Interfaces
Combined RDF
Model
Combined
KnowledgeModel
Application 1
Application 2
Application 3
Application
CSV
Parser
WEB
Parser
Application 2
Application 3
Semantic Bridges
Sensor
Network
Gateway
Server
Data
Source
Relational
DB
Relational
DB
RDB
Parser
CSV
Parser
WEB
Parser
RDF
Interfaces
Combined RDF
Model
Combined
Knowledge Model
Zoom
Application 1
Application 2
Application 3
Application
RDB
Parser
CSV
Parser
WEB
Parser
Model Knowledge Model
Application 1
Application 2
Application 3
er
RDB
Parser
CSV
Parser
WEB
Parser
RDF
Interfaces
Combined RDF
Model
Combined
KnowledgeModel
Application 1
Application 2
Application 3
Application
CSV
Parser
WEB
Parser
Application 2
Application 3
Data Integration
Data Sets
Combined RDF
Model
Combined
Knowledge Model
◦ Data from different sources is
combined into a common model
◦ The whole is greater than the sum
of its parts
◦ New knowledge can be obtained
Data Integration
Animal
MammalReptile
Human
Canine Feline
subClassOf subClassOf
subClassOf subClassOf
subClassOf
Wolves Terriers
Hounds
subClassOf
subClassOf
subClassOf
Foundation
Ontology
Extended
Ontology
◦ Foundation ontologies transcend
boundaries of single knowledge domain
◦ Common environment for
◦ Different terminologies
◦ Different knowledge domains
◦ Makes data integration easier
◦ Can be done (semi) automatically
◦ Easier to obtain new knowledge
M3 Framework
◦ Four data sources
◦ Different domains
◦ Overlapping data
◦ Same vocabulary
◦ Combined knowledge model
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Framework
◦ Smart Band sends a set of
measurements about user
◦ One of the measurements is
body temperature
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Graph View
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Framework
◦ Naturopathy expert describes
lemon and it’s properties
◦ Lemon is good to treat cold
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Graph View
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Framework
◦ Doctor describes High Fever as
symptom of Cold
◦ Given
◦ Doctor’s info
◦ Lemon’s properties
◦ Framework can infer that
◦ Lemon is good to treat High Fever
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Graph View
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Framework
◦ User creates a rule:
◦ If body temperature is higher than 38
◦ Then user has High Fever
◦ Given
◦ Sensor measurement
◦ User’s rule
◦ Doctor’s info
◦ Framework can infer that
◦ User has Cold
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Graph View
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Framework
◦ Given all the data
◦ Framework can recommend to
the user a lemon tea to treat the
cold
Adapted from: Machine-to-Machine Measurement (M3) Framework
M3 Graph View
Adapted from: Machine-to-Machine Measurement (M3) Framework
Linked Open World
◦ Linked Open Data
◦ Data repositories (DataHub, Data.gov, etc.)
◦ Share data to generate new data
◦ Linked Open Vocabularies
◦ Vocabularies repositories
◦ Facilitates data integration
◦ Linked Open Rules
◦ Rules repositories
◦ Concept only
◦ Linked Open Services
◦ Service repositories
◦ Concept only
Thanks for your attention

Semantic web: An overview

  • 1.
    Semantic Web PRESENTED BY:ANDRÉ MAZAYEV ANDRE.MAZ.90[ AT]GMAIL[DOT]COM
  • 2.
    Semantic Web? Whaaat? ◦What semantic web means? ◦ Smarter web!! Duuuh!
  • 3.
    Semantic Web? Whaaat? ◦What semantic web means? ◦ Smarter web!! Duuuh! ◦ Ok. But more specifically? ◦ It’s a web where it is easier to find stuff on internet
  • 4.
    Semantic Web? Whaaat? ◦What semantic web means? ◦ Smarter web!! Duuuh! ◦ Ok. But more specifically? ◦ It’s a web where it is easier to find stuff on internet ◦ Yeah! But how? ◦ Hmmmmm……
  • 5.
    Web 2.0 ◦ SearchProcess ◦ Refine search as you go ◦ The user is guiding the search accordingly to the results that are shown ◦ Search engine is only performing syntax based pattern match ◦ Plus some features to improve performance and accuracy ◦ Semantics are not used or used in a limited way during the search process
  • 6.
    Syntax and Semantics ◦Syntax ◦ About form ◦ Semantics ◦ About meaning
  • 7.
    Syntax and Semantics ◦Syntax ◦ Green, Yellow, Red ◦ Semantics ◦ Green = Go ◦ Yellow = Better stop ◦ Red = Stop Traffic Light Adapted from: Semantic Web from the 2013 Perspective
  • 8.
    User’s Web Example Exampleof dumb web ◦ Goal ◦ Find the telephone number of James Bond
  • 9.
    User’s Web Example Exampleof dumb web ◦ Goal ◦ Find the telephone number of James Bond ◦ For humans the answer is easy to find ◦ James Bond’s telephone number is 1-800-555-0199 ◦ James Bond is a fictional MI6 agent ◦ Since it’s a fictional agent we can infer that the number must be fake
  • 10.
    Machine’s Web Example Exampleof dumb web Source code of dumb web ◦ For machines find Bond’s number is a hard task ◦ No machine “readable” semantics ◦ Current Web ◦ Created for document sharing ◦ Instead of data sharing ◦ Adapted for Human to Human ◦ Machine to Machine communication is difficult
  • 11.
    Smart vs DumbWeb Example of dumb web Example of smart web
  • 12.
    Smart vs DumbWeb Visually both pages are identical Smart page carries much more “meaning” Example of dumb web Example of smart web
  • 13.
    Smart vs DumbWeb Source code of smart webSource code of dumb web
  • 14.
    Source code analysis Containsmore machine friendly structure ◦ Vocabulary is defined ◦ Data is structured ◦ Data is enriched The data can be represented as a graph Source code of smart web
  • 15.
  • 16.
    James Bond 1-800-555-0199 James Bond typeof name telephone Person Source codeanalysis With structured data it’s easy for a machine to find Bond’s telephone number Source code of smart web
  • 17.
    Graph analysis ◦ Simplestatements ◦ Subject – Predicate – Object ◦ All elements have their own URL ◦ Data is structured ◦ Data can be explored by machines James Bond 1-800-555-0199 James Bond typeof name telephone Person URL URL URL
  • 18.
    Structured Data Tool Sourcecode of a page with semantic markup Extracted data
  • 19.
    Structured Data Tool Extracteddata ◦ Data recognized by Google’s web crawler ◦ With structured data answers are easy to get ◦ What? ◦ Where? ◦ When Open?
  • 20.
    Semantic Web Present Future Webof Documents Web of Data Small Change Big Difference ◦ Data is explicit ◦ Data is connected ◦ Data can be explored by machines ◦ Nontrivial connections can be found ◦ Demo ◦ RelFinder
  • 21.
  • 22.
  • 23.
    RDF ◦ Resource DescriptionFramework ◦ Simple statements (triples) ◦ Subject – Predicate – Object ◦ Building block of RDFS and OWL ◦ Multiple serialization formats ◦ RDF/XML ◦ Turtle ◦ N-Triples Bond example in Turtle
  • 24.
  • 25.
    RDFS ◦ RDF Schema ◦Limited expressivity ◦ Describes classes, subclasses and properties ◦ Primary focused on “is a” and “sub class of” relationships Vocabulary Canine Animal Human Mammal Feline Reptile Taxonomy Animal MammalReptile Human Canine Feline subClassOf subClassOf subClassOf subClassOf subClassOf
  • 26.
  • 27.
    SPARQL ◦ SPARQL ProtocolAnd RDF Query Language ◦ SQL-Like structure James Bond 1-800-555-0199 James Bond typeof name telephone Person Graph
  • 28.
    SPARQL ◦ SPARQL ProtocolAnd RDF Query Language ◦ SQL-Like structure James Bond 1-800-555-0199 James Bond typeof name telephone Person Graph Goal: Find Bond’s Number
  • 29.
    SPARQL ◦ SPARQL ProtocolAnd RDF Query Language ◦ SQL-Like structure James Bond 1-800-555-0199 James Bond typeof name telephone Person Graph Query Goal: Find Bond’s Number
  • 30.
    SPARQL ◦ SPARQL ProtocolAnd RDF Query Language ◦ SQL-Like structure James Bond 1-800-555-0199 James Bond typeof name telephone Person Graph Answer Query Goal: Find Bond’s Number
  • 31.
  • 32.
    OWL ◦ Web OntologyLanguage ◦ Highly expressive ◦ Brings expressivity of logic to Semantic Web ◦ More expressive than RDFS ◦ Allows to express ◦ Constraints ◦ Cardinality ◦ Unions ◦ Intersections ◦ Etc. Resource that has property hasParent with value Bond belongs to a class named BondChild OWL Restriction Note: Often the concepts of taxonomies and ontologies overlap and used to describe same thing
  • 33.
  • 34.
    SWRL ◦ Semantic WebRule Language ◦ Combines parts from OWL and Datalog ◦ Rule syntax ◦ If body (antecedent) then assert head (consequent) x3 is x1’s uncle
  • 35.
    Under Development ◦ Pendingquestions ◦ How to ensure security of data? ◦ How to validate new data? ◦ Is source data reliable?
  • 36.
    Data Silos ◦ Eachapplication has its own ◦ Goals ◦ Vocabularies ◦ Knowledge base ◦ Not integrated with other data systems ◦ May have overlapping data Application 1 Application 2 Application 3 Sensor Network Gateway Server Application Data Source Relational DB Relational DB
  • 37.
    Semantic Bridges Sensor Network Gateway Server Data Source Relational DB Relational DB RDB Parser CSV Parser WEB Parser RDF Interfaces Combined RDF Model Combined KnowledgeModel Application 1 Application 2 Application 3 Application RDB Parser CSV Parser WEB Parser Model Knowledge Model Application 1 Application 2 Application 3 er RDB Parser CSV Parser WEB Parser RDF Interfaces Combined RDF Model Combined KnowledgeModel Application 1 Application 2 Application 3 Application CSV Parser WEB Parser Application 2 Application 3
  • 38.
    Semantic Bridges Sensor Network Gateway Server Data Source Relational DB Relational DB RDB Parser CSV Parser WEB Parser RDF Interfaces Combined RDF Model Combined KnowledgeModel Zoom Application 1 Application 2 Application 3 Application RDB Parser CSV Parser WEB Parser Model Knowledge Model Application 1 Application 2 Application 3 er RDB Parser CSV Parser WEB Parser RDF Interfaces Combined RDF Model Combined KnowledgeModel Application 1 Application 2 Application 3 Application CSV Parser WEB Parser Application 2 Application 3
  • 39.
    Data Integration Data Sets CombinedRDF Model Combined Knowledge Model ◦ Data from different sources is combined into a common model ◦ The whole is greater than the sum of its parts ◦ New knowledge can be obtained
  • 40.
    Data Integration Animal MammalReptile Human Canine Feline subClassOfsubClassOf subClassOf subClassOf subClassOf Wolves Terriers Hounds subClassOf subClassOf subClassOf Foundation Ontology Extended Ontology ◦ Foundation ontologies transcend boundaries of single knowledge domain ◦ Common environment for ◦ Different terminologies ◦ Different knowledge domains ◦ Makes data integration easier ◦ Can be done (semi) automatically ◦ Easier to obtain new knowledge
  • 41.
    M3 Framework ◦ Fourdata sources ◦ Different domains ◦ Overlapping data ◦ Same vocabulary ◦ Combined knowledge model Adapted from: Machine-to-Machine Measurement (M3) Framework
  • 42.
    M3 Framework ◦ SmartBand sends a set of measurements about user ◦ One of the measurements is body temperature Adapted from: Machine-to-Machine Measurement (M3) Framework
  • 43.
    M3 Graph View Adaptedfrom: Machine-to-Machine Measurement (M3) Framework
  • 44.
    M3 Framework ◦ Naturopathyexpert describes lemon and it’s properties ◦ Lemon is good to treat cold Adapted from: Machine-to-Machine Measurement (M3) Framework
  • 45.
    M3 Graph View Adaptedfrom: Machine-to-Machine Measurement (M3) Framework
  • 46.
    M3 Framework ◦ Doctordescribes High Fever as symptom of Cold ◦ Given ◦ Doctor’s info ◦ Lemon’s properties ◦ Framework can infer that ◦ Lemon is good to treat High Fever Adapted from: Machine-to-Machine Measurement (M3) Framework
  • 47.
    M3 Graph View Adaptedfrom: Machine-to-Machine Measurement (M3) Framework
  • 48.
    M3 Framework ◦ Usercreates a rule: ◦ If body temperature is higher than 38 ◦ Then user has High Fever ◦ Given ◦ Sensor measurement ◦ User’s rule ◦ Doctor’s info ◦ Framework can infer that ◦ User has Cold Adapted from: Machine-to-Machine Measurement (M3) Framework
  • 49.
    M3 Graph View Adaptedfrom: Machine-to-Machine Measurement (M3) Framework
  • 50.
    M3 Framework ◦ Givenall the data ◦ Framework can recommend to the user a lemon tea to treat the cold Adapted from: Machine-to-Machine Measurement (M3) Framework
  • 51.
    M3 Graph View Adaptedfrom: Machine-to-Machine Measurement (M3) Framework
  • 52.
    Linked Open World ◦Linked Open Data ◦ Data repositories (DataHub, Data.gov, etc.) ◦ Share data to generate new data ◦ Linked Open Vocabularies ◦ Vocabularies repositories ◦ Facilitates data integration ◦ Linked Open Rules ◦ Rules repositories ◦ Concept only ◦ Linked Open Services ◦ Service repositories ◦ Concept only
  • 53.
    Thanks for yourattention