The GoodRelations Ontology
       for E-Commerce

          3rd KRDB School on
Trends in the Web of Data (KRDB-2010)
       Brixen-Bressanone, Italy,
        17-18 September 2010

          Prof. Dr. Martin Hepp
Professur für Allgemeine BWL, insbesondere E-Business
Part 1: Why bother?




18.09.2010                         2
6. Upcoming Research Challenges
Part 1: Why bother?




18.09.2010                         4
Matchmaking in Market Economies




18.09.2010                             5
Macroeconomic Impact

                   Transaction Costs:
                     > 50 % of the
                     US GDP (1970)
             John Joseph Wallis and Douglas C. North:
              Measuring the Transaction Sector in the
                 American Economy, 1870 – 1970
                              (1986)
18.09.2010                                              6
Key Driver of Search Costs:
           Specificity

How much you loose when you can‘t
use a good for what it was designed.
Growth in Specificity




             1920: 5168 Types of Goods

18.09.2010
                                         8
Examples 2010




18.09.2010                   9
Examples 2010




18.09.2010                   10
Examples 2010




18.09.2010                   11
Specificity Increases the
                 Search Space




18.09.2010                               12
WWW:
    Dramatic Reduction of Search Effort




             1993            2010
   Lower search costs per search than
          ever before in history.
18.09.2010                                13
But ….
The WWW: A Giant Data Shredder




     Source:                   Recipient:
 Structured Data            Unstructured Text



18.09.2010                              15
What is Linked Data Linked
                                 loves


                          Susi           Martin



       1            2              3              4




18.09.2010                                            16
What is Special About E-Commerce Data?
 1

 2           RDBMS



 3

 4           $$$
18.09.2010                        17
GoodRelations: A Global Schema for
    Commerce Data on the Web
                                            Extraction
                        Arbitrary Query     and Reuse


Manufacturers
                                                     Retailers
Payment
                                                     Delivery
Product Model                                 Warranty
 Master Data     Shop                Spare Parts &
                Offerings   Auctions Consumables
18.09.2010                                               18
On the Shoulders of Giants




     A Unified View of Commerce Data
                on the Web
18.09.2010                                19
GoodRelations Deployment: Small Data
      Packets Inside Your Page (RDFa)




18.09.2010                                 20
Valuable Types of Links:
             Product - Product Model




                                                             Photo credits: Flickr.com, available
                                                              under CC BY 2.0 by bsabarnowl
   Ford T
   Data-      gr:hasMakeAndModel
   sheet




       Often via strong, non-URI identifiers like EAN/UPC
18.09.2010                                                  21
Valuable Types of Links:
                 Offer – Store(s)


    XYZ
              gr:availableAtOrFrom
    for $
     99




18.09.2010                              22
Valuable Types of Links:
              Company – Store(s)



                      gr:hasPOS




18.09.2010                              23
Part 2: Ontology Engineering
                       Revisited




18.09.2010                                  24
Immanuel Kant on Ontologies
                   & Linked Data
               „Thoughts without content are empty,
               intuitions without concepts are blind.“
                                 Critique of Pure Reason (1781)


  1. Ontologies without data are useless
  2. Data without ontologies is blind


18.09.2010                                                 25
In other words: Schemas Matter




                                             Photo credits: Flickr.com, available
                                               under CC BY 2.0 by dnorman
    Otherwise your data is just landfill…
18.09.2010                                  26
Albert Einstein on Schema Design

"Make everything as simple as possible, but
                not simpler.“
                              Albert Einstein




18.09.2010                                 27
Data, Standards, Ontologies




18.09.2010                                 28
Subtle Distinctions Foster Data Reuse
• Product      Offer
     – „You can buy or lease my house“
• Store      Business entity
     – „How many Tesco stores are in London?“
• Product      Product Model
     – „How many digital cameras by Canon are
       listed on eBay“?

18.09.2010                                      29
Sophisticated Category Systems:

         Foundation for Intelligence and
                  Judgment




18.09.2010                                 30
18.09.2010
                                                            Ontology Economics




  Hepp, Martin: Possible Ontologies: How Reality Constrains the
  Development of Relevant Ontologies, in: IEEE Internet Computing,
31




  Vol. 11, No. 1, pp. 90-96, Jan-Feb 2007
Incremental Granularity
               & Lexical Carry-Over




18.09.2010                             32
Ontology Engineering
• Generic model
     – Stable distinctions
     – Easy to populate
     – Incremental Enrichment
• Good textual elements
• Good documentation
• Tool support for the entire tool chain

18.09.2010                                 33
Part 3: GoodRelations Overview




18.09.2010                               34
Basic Structure of Offers:
Agent-Promise-Object Principle
                                         Object or
Agent 1             Promise
                                         Happening

          Compensation     Transfer of
                             Rights




                     Agent 2



                                                     35
The Minimal Scenario
• Scope
  –   Business entity
  –   Points-of-sale
  –   Opening hours
  –   Payment options
• Suitable for
  – Every business
  – E-commerce and
    brick-and-mortar

                                    36
The Simple Scenario
• Scope: Minimal scenario plus
  – Range of products or services
  – Business functions
  – Eligible regions or customer
    types
  – Delivery options
• Suitable for
  – Any business: E-Commerce and
    brick-and-mortar
  – Specific products or services
                                    37
The Comprehensive Scenario
• Scope: Simple scenario plus
   –   Individual products or services
   –   Product features
   –   Pricing, rebates, etc.
   –   Availability
• Suitable for
   – Any business: E-commerce and
     brick-and-mortar
   – Specific products or services
   – Structured product database


                                         38
Product Model Data Scenario
• Scope
  – Individual product
    models
  – Quantitative and
    qualitative features
• Suitable for
  – Manufacturers of
    commodities



                                    39
Developer Resources, Data, Tools




   http://purl.org/goodrelations/



18.09.2010                              40
The Minimal Scenario (UML & RDF/N3)




18.09.2010                        41
The Simple Scenario: UML




18.09.2010                              42
The Simple Scenario: RDF/N3 - Details




18.09.2010                          43
Alternative Ways of Describing the
              Product or Service
• Omit it
     – Minimal Example: Describe just your business & store
• gr:ProductOrServiceSomeInstancesPlaceholder + rdfs:comment
     – Textual
• Product or service ontology
     – eclassOWL
     – freeClass
• DBPedia URIs
• Turn proprietary hierarchy into pseudo-ontology

18.09.2010                                                    44
Impact and Success
• One of the few vocabularies implemented
  by major businesses out of their own
  budgets
• BestBuy, O’Reilly, Overstock.com,…
• Ca. 16 % of all triples as of now
• Supported by Yahoo
• Bing, Google may join

18.09.2010                                  45
Yahoo Enhanced by SearchMonkey




18.09.2010                           46
Incredible Success




18.09.2010                        47
GoodRelations #2 of all Web Ontologies




         …and this does not yet include the > 10 Mio. offers
         from Amazon and eBay!

18.09.2010                                                     48
GoodRelations #2 of all Web Ontologies




18.09.2010                          49
GoodRelations Design Principles
• Keep simple things       Lightweight         Heavyweight
  simple and make          Web of Data         Web of Data
  complex things
  possible                    LOD                OWL DL
• Cater for LOD and OWL   RDF + a little bit
  DL worlds
• Academically sound
• Industry-strength
  engineering
• Practically relevant

18.09.2010                                            50
Syntax-neutral
•   RDF/XML, Turtle         • Microdata
•   RDFa                    • dataRSS
•   OData
•   GData




http://www.ebusiness-unibw.org/wiki/Syntaxes4GoodRelations
18.09.2010                                            51
Part 4: Publishing GoodRelations Data




18.09.2010                            52
RDFa in Snippet Style




    http://www.ebusiness-unibw.org/tools/rdf2rdfa/
18.09.2010                                           53
Publishing GoodRelations Data
• RDFa in Snippet Style
• sitemap.xml with proper lastmod attribute
• robots.txt




18.09.2010                               54
Microdata in Snippet Style




 http://www.ebusiness-unibw.org/tools/rdf2microdata/
18.09.2010                                      55
Part 5: GoodRelations Advanced
                   Topics




18.09.2010                              56
GoodRelations-compliant Domain Ontologies




18.09.2010                                       57
Meta-Model for Quantitative Data




18.09.2010                                58
Both Sides Can Help Build a Bridge


    gr:seeks property




18.09.2010                               59
Ownership & Self Exposure
• gr:owns property




18.09.2010                               60
6. Upcoming Research Challenges
Research Challenges
(1)    Natural Language Processing
(2)    Ontology Mapping and Alignment
(3)    Collaborative Ontology Engineering
(4)    Crawling, Update, Federation
(5)    Matchmaking & Query Learning
(6)    Applications and Interaction Patterns
(7)    Storage and Reasoning
18.09.2010                                     62
Natural Language Processing




18.09.2010                                 63
Ontology Mapping and Alignment




18.09.2010                              64
Collaborative Ontology Engineering
• OpenVocab
• Knoodl
• Protégé Collaboration
  Support
• OntoVerse
• MyOntology
• Twine Ontology Editor
• Neologism
• MoKi
http://www.ebusiness-unibw.org/wiki/Own_GoodRelations_Vocabularies
18.09.2010                                                     65
Crawling, Update, Federation
(1) Shop data changes every 1..24 h
(2) Can you harvest the data from 1,000,000
   shop sites just via
    – Sitemap.xml with proper lastmod
      attribute
    – RDFa inside the pages



18.09.2010                                  66
Matchmaking & Query Learning




18.09.2010                              67
Applications and Interaction Patterns




18.09.2010                                 68
Storage and Reasoning
• RDFS-style reasoning
• Non-standard inference rules
• Massive scale
     – 1 Mio shops etc.
     – 1 k – 100 k items,let’s say 10 k
     – 100 triples per item
     – 1 Mio * 10 k * 100 = 1,000,000,000,000
     – 1 trillion triples
18.09.2010                                      69
Storage and Reasoning
• Hybrid queries




18.09.2010                           70
Data Quality Management




http://www.ebusiness-unibw.org/tools/goodrelations-validator/
18.09.2010                                                71
Thank you!
                       http://purl.org/goodrelations/

                         Prof. Dr. Martin Hepp
             Chair of General Management and E-Business
                 Universitaet der Bundeswehr Muenchen
                      Werner-Heisenberg-Weg 39
                       D-85579 Neubiberg, Germany
                       Phone: +49 89 6004-4217
                          Fax: +49 89 6004-4620
                   http://www.unibw.de/ebusiness/

             http://purl.org/goodrelations/
                         mhepp@computer.org
18.09.2010                                                72

KRDB2010-GoodRelations

  • 1.
    The GoodRelations Ontology for E-Commerce 3rd KRDB School on Trends in the Web of Data (KRDB-2010) Brixen-Bressanone, Italy, 17-18 September 2010 Prof. Dr. Martin Hepp Professur für Allgemeine BWL, insbesondere E-Business
  • 2.
    Part 1: Whybother? 18.09.2010 2
  • 3.
  • 4.
    Part 1: Whybother? 18.09.2010 4
  • 5.
    Matchmaking in MarketEconomies 18.09.2010 5
  • 6.
    Macroeconomic Impact Transaction Costs: > 50 % of the US GDP (1970) John Joseph Wallis and Douglas C. North: Measuring the Transaction Sector in the American Economy, 1870 – 1970 (1986) 18.09.2010 6
  • 7.
    Key Driver ofSearch Costs: Specificity How much you loose when you can‘t use a good for what it was designed.
  • 8.
    Growth in Specificity 1920: 5168 Types of Goods 18.09.2010 8
  • 9.
  • 10.
  • 11.
  • 12.
    Specificity Increases the Search Space 18.09.2010 12
  • 13.
    WWW: Dramatic Reduction of Search Effort 1993 2010 Lower search costs per search than ever before in history. 18.09.2010 13
  • 14.
  • 15.
    The WWW: AGiant Data Shredder Source: Recipient: Structured Data Unstructured Text 18.09.2010 15
  • 16.
    What is LinkedData Linked loves Susi Martin 1 2 3 4 18.09.2010 16
  • 17.
    What is SpecialAbout E-Commerce Data? 1 2 RDBMS 3 4 $$$ 18.09.2010 17
  • 18.
    GoodRelations: A GlobalSchema for Commerce Data on the Web Extraction Arbitrary Query and Reuse Manufacturers Retailers Payment Delivery Product Model Warranty Master Data Shop Spare Parts & Offerings Auctions Consumables 18.09.2010 18
  • 19.
    On the Shouldersof Giants A Unified View of Commerce Data on the Web 18.09.2010 19
  • 20.
    GoodRelations Deployment: SmallData Packets Inside Your Page (RDFa) 18.09.2010 20
  • 21.
    Valuable Types ofLinks: Product - Product Model Photo credits: Flickr.com, available under CC BY 2.0 by bsabarnowl Ford T Data- gr:hasMakeAndModel sheet Often via strong, non-URI identifiers like EAN/UPC 18.09.2010 21
  • 22.
    Valuable Types ofLinks: Offer – Store(s) XYZ gr:availableAtOrFrom for $ 99 18.09.2010 22
  • 23.
    Valuable Types ofLinks: Company – Store(s) gr:hasPOS 18.09.2010 23
  • 24.
    Part 2: OntologyEngineering Revisited 18.09.2010 24
  • 25.
    Immanuel Kant onOntologies & Linked Data „Thoughts without content are empty, intuitions without concepts are blind.“ Critique of Pure Reason (1781) 1. Ontologies without data are useless 2. Data without ontologies is blind 18.09.2010 25
  • 26.
    In other words:Schemas Matter Photo credits: Flickr.com, available under CC BY 2.0 by dnorman Otherwise your data is just landfill… 18.09.2010 26
  • 27.
    Albert Einstein onSchema Design "Make everything as simple as possible, but not simpler.“ Albert Einstein 18.09.2010 27
  • 28.
  • 29.
    Subtle Distinctions FosterData Reuse • Product Offer – „You can buy or lease my house“ • Store Business entity – „How many Tesco stores are in London?“ • Product Product Model – „How many digital cameras by Canon are listed on eBay“? 18.09.2010 29
  • 30.
    Sophisticated Category Systems: Foundation for Intelligence and Judgment 18.09.2010 30
  • 31.
    18.09.2010 Ontology Economics Hepp, Martin: Possible Ontologies: How Reality Constrains the Development of Relevant Ontologies, in: IEEE Internet Computing, 31 Vol. 11, No. 1, pp. 90-96, Jan-Feb 2007
  • 32.
    Incremental Granularity & Lexical Carry-Over 18.09.2010 32
  • 33.
    Ontology Engineering • Genericmodel – Stable distinctions – Easy to populate – Incremental Enrichment • Good textual elements • Good documentation • Tool support for the entire tool chain 18.09.2010 33
  • 34.
    Part 3: GoodRelationsOverview 18.09.2010 34
  • 35.
    Basic Structure ofOffers: Agent-Promise-Object Principle Object or Agent 1 Promise Happening Compensation Transfer of Rights Agent 2 35
  • 36.
    The Minimal Scenario •Scope – Business entity – Points-of-sale – Opening hours – Payment options • Suitable for – Every business – E-commerce and brick-and-mortar 36
  • 37.
    The Simple Scenario •Scope: Minimal scenario plus – Range of products or services – Business functions – Eligible regions or customer types – Delivery options • Suitable for – Any business: E-Commerce and brick-and-mortar – Specific products or services 37
  • 38.
    The Comprehensive Scenario •Scope: Simple scenario plus – Individual products or services – Product features – Pricing, rebates, etc. – Availability • Suitable for – Any business: E-commerce and brick-and-mortar – Specific products or services – Structured product database 38
  • 39.
    Product Model DataScenario • Scope – Individual product models – Quantitative and qualitative features • Suitable for – Manufacturers of commodities 39
  • 40.
    Developer Resources, Data,Tools http://purl.org/goodrelations/ 18.09.2010 40
  • 41.
    The Minimal Scenario(UML & RDF/N3) 18.09.2010 41
  • 42.
    The Simple Scenario:UML 18.09.2010 42
  • 43.
    The Simple Scenario:RDF/N3 - Details 18.09.2010 43
  • 44.
    Alternative Ways ofDescribing the Product or Service • Omit it – Minimal Example: Describe just your business & store • gr:ProductOrServiceSomeInstancesPlaceholder + rdfs:comment – Textual • Product or service ontology – eclassOWL – freeClass • DBPedia URIs • Turn proprietary hierarchy into pseudo-ontology 18.09.2010 44
  • 45.
    Impact and Success •One of the few vocabularies implemented by major businesses out of their own budgets • BestBuy, O’Reilly, Overstock.com,… • Ca. 16 % of all triples as of now • Supported by Yahoo • Bing, Google may join 18.09.2010 45
  • 46.
    Yahoo Enhanced bySearchMonkey 18.09.2010 46
  • 47.
  • 48.
    GoodRelations #2 ofall Web Ontologies …and this does not yet include the > 10 Mio. offers from Amazon and eBay! 18.09.2010 48
  • 49.
    GoodRelations #2 ofall Web Ontologies 18.09.2010 49
  • 50.
    GoodRelations Design Principles •Keep simple things Lightweight Heavyweight simple and make Web of Data Web of Data complex things possible LOD OWL DL • Cater for LOD and OWL RDF + a little bit DL worlds • Academically sound • Industry-strength engineering • Practically relevant 18.09.2010 50
  • 51.
    Syntax-neutral • RDF/XML, Turtle • Microdata • RDFa • dataRSS • OData • GData http://www.ebusiness-unibw.org/wiki/Syntaxes4GoodRelations 18.09.2010 51
  • 52.
    Part 4: PublishingGoodRelations Data 18.09.2010 52
  • 53.
    RDFa in SnippetStyle http://www.ebusiness-unibw.org/tools/rdf2rdfa/ 18.09.2010 53
  • 54.
    Publishing GoodRelations Data •RDFa in Snippet Style • sitemap.xml with proper lastmod attribute • robots.txt 18.09.2010 54
  • 55.
    Microdata in SnippetStyle http://www.ebusiness-unibw.org/tools/rdf2microdata/ 18.09.2010 55
  • 56.
    Part 5: GoodRelationsAdvanced Topics 18.09.2010 56
  • 57.
  • 58.
    Meta-Model for QuantitativeData 18.09.2010 58
  • 59.
    Both Sides CanHelp Build a Bridge gr:seeks property 18.09.2010 59
  • 60.
    Ownership & SelfExposure • gr:owns property 18.09.2010 60
  • 61.
  • 62.
    Research Challenges (1) Natural Language Processing (2) Ontology Mapping and Alignment (3) Collaborative Ontology Engineering (4) Crawling, Update, Federation (5) Matchmaking & Query Learning (6) Applications and Interaction Patterns (7) Storage and Reasoning 18.09.2010 62
  • 63.
  • 64.
    Ontology Mapping andAlignment 18.09.2010 64
  • 65.
    Collaborative Ontology Engineering •OpenVocab • Knoodl • Protégé Collaboration Support • OntoVerse • MyOntology • Twine Ontology Editor • Neologism • MoKi http://www.ebusiness-unibw.org/wiki/Own_GoodRelations_Vocabularies 18.09.2010 65
  • 66.
    Crawling, Update, Federation (1)Shop data changes every 1..24 h (2) Can you harvest the data from 1,000,000 shop sites just via – Sitemap.xml with proper lastmod attribute – RDFa inside the pages 18.09.2010 66
  • 67.
    Matchmaking & QueryLearning 18.09.2010 67
  • 68.
    Applications and InteractionPatterns 18.09.2010 68
  • 69.
    Storage and Reasoning •RDFS-style reasoning • Non-standard inference rules • Massive scale – 1 Mio shops etc. – 1 k – 100 k items,let’s say 10 k – 100 triples per item – 1 Mio * 10 k * 100 = 1,000,000,000,000 – 1 trillion triples 18.09.2010 69
  • 70.
    Storage and Reasoning •Hybrid queries 18.09.2010 70
  • 71.
  • 72.
    Thank you! http://purl.org/goodrelations/ Prof. Dr. Martin Hepp Chair of General Management and E-Business Universitaet der Bundeswehr Muenchen Werner-Heisenberg-Weg 39 D-85579 Neubiberg, Germany Phone: +49 89 6004-4217 Fax: +49 89 6004-4620 http://www.unibw.de/ebusiness/ http://purl.org/goodrelations/ mhepp@computer.org 18.09.2010 72