luk vervenne
synergetics nv   1
Een veelkoppig monster

• Individu als “de nieuwe stakeholder in zijn eigen
  processen” (Life Management)
• Van syntax naar semantiek
• Markten, jobs evolueren steeds sneller
• Competenties: duizend bloemen bloeien
• Top-down vs bottom-up
• …

                                                      2
Labour Markets
 beyond the client/server paradigm

Regional Employability Ecosystems
        Server         Server         Server
                                                   Industry / Sector specific
                                                   Processes & Services
        CORP.           GOV.          Service
                                     Provider
                                                   Base Infrastructure for
     Personal InfrastructureClient
        Client    Client      Cloud                Region & Sector-wide
                                                   User-centric / User-driven
                                                   Ecosystems
                                                      1.   Personal Infrastructure
Organising the communality                            2.   Semantic Coordination
Assure & Promote Labour Market Mobility:              3.   BI Infrastructure
1.   Governance (PPP)                                 4.   eContent gateway
2.   Communality Based Infrastructure & Services      5.   Matching Infrastructure
3.   Semantics                                        6.   Trust & Security (+TTP)
4.   Trust & Security
1. Bouwen op wat we hebben

       (NEN norm)




                             4
Luk Vervenne – Dr. Ingo Dahn
Application Profiles
• Standards are too general – I don’t need all that fuzz




• Standards are too restricted – they don’t let me do
  what I want


• Solution: Application Profiles!
Make My Day!
           Make My Profile – it‘s easy!

•   Making mandatory what I do want
•   Making optional what I tolerate
•   Remove what I don’t need
•   Add what I need

• Next step:
  • Convince your industry sector
  • Agree
  • Share
Making My Profile – Oh so tricky!

• My own extensions
• Mixing and matching many profiles
• All referenced files must :
   • exist
   • validate against another profile or one of a few…
• And…
   • you have to find out which one to use…
   • the imsmanifest.xml must exist !!
Prerequisites of Profiles
• A community of stakeholders

• Acquaintance with:
   • their needs…
   • their willingness to agree…
   • their willingness to implement!!!


• STEPS: Success of application profile depends on
  implementers, data providers and data consumers
Conformance Testing
• Test so that data conforms to YOUR profile
• Problem:
   • Each profile requires a specific test system
   • Test system development is expensive
• Solution:
   • Capture profile in machine readable form
   • Configure generic test system
Profiling: SchemaProf
Creating a   Test Service
Running the Tests
Survey Report
Detailed Reports
2. En wat met niet-gestructureerde data?

       (competentiebeschrijvingen
          beroepsbeschrijvingen
               ervaringen
               vacatures
                   …)


                                           16
Purpose:
   Semantic Comparison of Labour Market Data

Compare real-world employability & employment data versus
Reference Data of Competences, Occupations, Qualifications, …

Allows the meaningful search, assessment or match of
experience, professional activities, skills & competences by using :
• Domain Semantics (Annotated Reference framework data)
• Linguistic Semantics (Unstructured data using NL processing)
    • Created individually (personal employability data)
    • Created at company level (vacancies, job profiles, …)

                                                                17
Approach

• Knowledge management
  • Knowledge encoding (knowledge bases)
  • Knowledge-based data processing
     •   Annotation
     •   Comparison
     •   Interoperation
     •   Inference
• (Natural Language) Data processing
  • Interpret data dynamically
  • Capture data individuality and specifics

                                               18
Data, Knowledge & Semantics
1. Data : experience, goal, competence, preference, hobby, training, job, task
2. Knowledge : frameworks, expert rules, models, ontology
3. Semantics : compare Data + Knowledge for semantic operations:
                 •   Data management
                 •   Knowledge management
                 •   Knowledge-based data management
                 •   Data-oriented knowledge management
Two kinds of Semantics are involved:
Precompiled: Static, knowledge-based operation
        = knowledge semantics
Extracted :      Dynamic, in real-time in data management
        = data semantics
                                                                         19
PROCES          PERSONAL
 CENTRIC       EMPLOYABILITY
PERSONAL   (SEMANTIC META-)DATA
  DATA        (Content + VOCs)

  GOV



  EDU


                         PDS
COMPANY


 SERVICE
PROVIDER



                e-          HR-
            PORTFOLIOS   PROCESSES

                     LEER-
                    DOSSIER
                                     20
Personal Data Infrastructure
                                                            New M a r k e t s
Linked-             NL                HR-XML         Social                       Personal
   In                                 GermanCV,                                   Finance
                 Portfolio            iProfile UK
                                                    Network
                                                                   Mobile           Data
                                                     Data
                                                                   context
                   eGov                                             data          Personal
EuroCV                               EuroPass       Consumer
               Citizen data                           Data
                                                                                   Health
                                                                                    Data



                        Import
                                                       1
                                         TRANS-
 ePortfolio                            FORMATION    CRUD WS
                         Export          engine
                                                      10%




                                                                    SOA GATEWAY
                                                        2
               Create, Read,             Web
                                                    CRUD WS
              Update, Delete           interface
                                                      20%                         PDS
                               Integration with
                                                       3
                                                      WS
                               (legacy) systems
                                                      70%
22
PROCES          PERSONAL                   SHARED                   (PARTIAL)
 CENTRIC       EMPLOYABILITY               SEMANTIC                 REFERENCE
PERSONAL   (SEMANTIC META-)DATA           KNOWLEDGE                FRAMEWORKS
  DATA        (Content + VOCs)               BASE
                                                                     CINOP/
  GOV
                                                                     ECVET
                                                                      UWV
  EDU
                                          SEMANTIC                    SHL
                                         KNOWLEDGE
                         PDS                BASE                      SBB
COMPANY
                                                                      VDAB
 SERVICE
                                                                      ESCO
PROVIDER      UNIFIED SEMANTIC MATCHING



                e-          HR-                        SKILLS &
                                     EVIDENCE        COMPETENCES
            PORTFOLIOS   PROCESSES

                     LEER-                      VACANCIES
                    DOSSIER
                                                                         23
The European need for
semantic interoperability




                            24
Beyond Asset Descriptions




                 FEDERATION




                              25
: Data & Semantics
                                        The Employability data from :
                                        (1) Candidates (employabilityPortfolio™, CVs,
  VACANCY            CANDIDATE          EuroPASS, HR, educational, Recruiting or Public or
                                        Private Employment Systems
                                        (2) Vacancies are indexed & correlated against
                                        vocational/occupational qualification and
                                        competency catalogues.
 ROME : Adult      COLO : Graduate
OCCUPATIONAL        VOCATIONAL          This demo is based on the French ROME 2.0
 Classification     Qualifications      catalogue. However such National catalogues
                                        (NOS/SSC) can be enriched with information from
                                        other European catalogues.
   ‘ESCO’ skills & competencies
          & occupations                 All vocational categories and qualifications and their
                                        respective competency / skills profiles are correlated.


                                        A competency profile is required for the performant
             ACTIVITY
           ABSTRACTIONS                 execution of professional/vocational ”activities”


      ONTOLOGY consists of
      BEHAVIORAL / SKILLS               Each activity is abstracted using concepts and their
          SEMANTICS                     relations
                                                     ©Synergetics 2012 All rights reserved
Activity Semantics
                                                 1 Job Category
• Competency framework (ROME CATALOG)            2 Associated competencies
• Semantic Activity abstractions     click       3 Semantic entities & relations


                                        2




            1

                                             3
Competency ANALOGY: similarities
Activity Semantics       Compare similarities between 2 competencies
• Job categories
• Competencies
  belonging to a
  job category




 Score on 1
  Common
  Relation




                           ©Synergetics 2010 all rights resverved
Competency GAP analysis : differencies
                                  Compare differences between 2 competencies
Activity Semantics
• Job categories
• Competencies
  belonging to a
  job category




 Score on 1
  Common
  Relation




                            ©Synergetics 2010 all rights resverved
Find candidates
                  Specify vacancies

1. Select ranking threshold
                                           2. Select (in)experienced candidates




                               3. Select the job offer to start the search




                ©Synergetics 2010 All rights reserved
Find candidates               All animals are equal!


                                    Click to visualise the competency

        Results Overview
        Semantic Ranking
 according to “presence”
of relevant competencies
Visualise the profile

                1. Click the ‘bar icon’, multiple selections are allowed




                                                                 3. Select graph type



2. Graphic representation of the “activity semantics” of the competency (indicators)

                            ©Synergetics 2010 all rights resverved
Find your candidate
                 Visualise the semantic differences of the
              7 competencies from the 9 selected candidates

                                                                  7 competencies
   …but some are more equal than others 




9 selected candidates



                          ©Synergetics 2010 All rights reserved
Conclusion
• The “Activity Semantics” extraction method makes use of Natural Language
  Technology :
   • Abstraction : Extract universal semantics from different competenvy
      descriptions
   • Interpretation : Skills and Competency extraction from
      employabilityPortfolios™, texts, CVs, vacancies and … Regional
      Employability Platform enabled systems!
• For this demo we used (anonymised) real life data from 13.000 candidates
  and 13.000 vacancies.
• The Activity Semantics method is based on its own ABAS Ontology
• Semantic annotation & enrichment of existing ESCO and other European
  occupation, vocation, qualification and competency frameworks (ROME,
  COLO, NOS, UKCES, SSC, QCF, …etc)
• This semantic method allows for considerable refined decision making when
  searching for the right candidate or vacancy, based upon a set of required or
  desired competencies.
                          ©Synergetics 2010 All rights reserved
Semantic DNA

The Technology
• Semantic metadata
   • Interpretation
       • Input:    Textual data
       • Output:   Semantic DNA
   • Comparison
       • Input:    Semantic DNAs
• Output: Scores of Similarity, Difference, Equivalence.
  These are the basis for semantic matching
• Robust text understanding technology
   • Language parsing and interpretation
   • Customisable and optimisable
   • Languages (French, English currently, Dutch coming up)
Use of Competence DNA
• Operation
  • Extraction of competence semantics
  • Semantic comparison of competences


• Customization by competence frameworks
  • Knowledge bases of competence frameworks
  • Language capability, based on knowledge bases


• Comparison of competences:
  • free2ref
  • free2free
Example free2free comparison

                   Text Editor for Input




  Text Input


                                     Semantic DNA




               Semantic comparison

10052012 luc vervenne synergetics van syntax portfolio naar semantische uitwisselbaarheid

  • 1.
  • 2.
    Een veelkoppig monster •Individu als “de nieuwe stakeholder in zijn eigen processen” (Life Management) • Van syntax naar semantiek • Markten, jobs evolueren steeds sneller • Competenties: duizend bloemen bloeien • Top-down vs bottom-up • … 2
  • 3.
    Labour Markets beyondthe client/server paradigm Regional Employability Ecosystems Server Server Server Industry / Sector specific Processes & Services CORP. GOV. Service Provider Base Infrastructure for Personal InfrastructureClient Client Client Cloud Region & Sector-wide User-centric / User-driven Ecosystems 1. Personal Infrastructure Organising the communality 2. Semantic Coordination Assure & Promote Labour Market Mobility: 3. BI Infrastructure 1. Governance (PPP) 4. eContent gateway 2. Communality Based Infrastructure & Services 5. Matching Infrastructure 3. Semantics 6. Trust & Security (+TTP) 4. Trust & Security
  • 4.
    1. Bouwen opwat we hebben (NEN norm) 4
  • 5.
    Luk Vervenne –Dr. Ingo Dahn
  • 6.
    Application Profiles • Standardsare too general – I don’t need all that fuzz • Standards are too restricted – they don’t let me do what I want • Solution: Application Profiles!
  • 7.
    Make My Day! Make My Profile – it‘s easy! • Making mandatory what I do want • Making optional what I tolerate • Remove what I don’t need • Add what I need • Next step: • Convince your industry sector • Agree • Share
  • 8.
    Making My Profile– Oh so tricky! • My own extensions • Mixing and matching many profiles • All referenced files must : • exist • validate against another profile or one of a few… • And… • you have to find out which one to use… • the imsmanifest.xml must exist !!
  • 9.
    Prerequisites of Profiles •A community of stakeholders • Acquaintance with: • their needs… • their willingness to agree… • their willingness to implement!!! • STEPS: Success of application profile depends on implementers, data providers and data consumers
  • 10.
    Conformance Testing • Testso that data conforms to YOUR profile • Problem: • Each profile requires a specific test system • Test system development is expensive • Solution: • Capture profile in machine readable form • Configure generic test system
  • 11.
  • 12.
    Creating a Test Service
  • 13.
  • 14.
  • 15.
  • 16.
    2. En watmet niet-gestructureerde data? (competentiebeschrijvingen beroepsbeschrijvingen ervaringen vacatures …) 16
  • 17.
    Purpose: Semantic Comparison of Labour Market Data Compare real-world employability & employment data versus Reference Data of Competences, Occupations, Qualifications, … Allows the meaningful search, assessment or match of experience, professional activities, skills & competences by using : • Domain Semantics (Annotated Reference framework data) • Linguistic Semantics (Unstructured data using NL processing) • Created individually (personal employability data) • Created at company level (vacancies, job profiles, …) 17
  • 18.
    Approach • Knowledge management • Knowledge encoding (knowledge bases) • Knowledge-based data processing • Annotation • Comparison • Interoperation • Inference • (Natural Language) Data processing • Interpret data dynamically • Capture data individuality and specifics 18
  • 19.
    Data, Knowledge &Semantics 1. Data : experience, goal, competence, preference, hobby, training, job, task 2. Knowledge : frameworks, expert rules, models, ontology 3. Semantics : compare Data + Knowledge for semantic operations: • Data management • Knowledge management • Knowledge-based data management • Data-oriented knowledge management Two kinds of Semantics are involved: Precompiled: Static, knowledge-based operation = knowledge semantics Extracted : Dynamic, in real-time in data management = data semantics 19
  • 20.
    PROCES PERSONAL CENTRIC EMPLOYABILITY PERSONAL (SEMANTIC META-)DATA DATA (Content + VOCs) GOV EDU PDS COMPANY SERVICE PROVIDER e- HR- PORTFOLIOS PROCESSES LEER- DOSSIER 20
  • 21.
    Personal Data Infrastructure New M a r k e t s Linked- NL HR-XML Social Personal In GermanCV, Finance Portfolio iProfile UK Network Mobile Data Data context eGov data Personal EuroCV EuroPass Consumer Citizen data Data Health Data Import 1 TRANS- ePortfolio FORMATION CRUD WS Export engine 10% SOA GATEWAY 2 Create, Read, Web CRUD WS Update, Delete interface 20% PDS Integration with 3 WS (legacy) systems 70%
  • 22.
  • 23.
    PROCES PERSONAL SHARED (PARTIAL) CENTRIC EMPLOYABILITY SEMANTIC REFERENCE PERSONAL (SEMANTIC META-)DATA KNOWLEDGE FRAMEWORKS DATA (Content + VOCs) BASE CINOP/ GOV ECVET UWV EDU SEMANTIC SHL KNOWLEDGE PDS BASE SBB COMPANY VDAB SERVICE ESCO PROVIDER UNIFIED SEMANTIC MATCHING e- HR- SKILLS & EVIDENCE COMPETENCES PORTFOLIOS PROCESSES LEER- VACANCIES DOSSIER 23
  • 24.
    The European needfor semantic interoperability 24
  • 25.
  • 26.
    : Data &Semantics The Employability data from : (1) Candidates (employabilityPortfolio™, CVs, VACANCY CANDIDATE EuroPASS, HR, educational, Recruiting or Public or Private Employment Systems (2) Vacancies are indexed & correlated against vocational/occupational qualification and competency catalogues. ROME : Adult COLO : Graduate OCCUPATIONAL VOCATIONAL This demo is based on the French ROME 2.0 Classification Qualifications catalogue. However such National catalogues (NOS/SSC) can be enriched with information from other European catalogues. ‘ESCO’ skills & competencies & occupations All vocational categories and qualifications and their respective competency / skills profiles are correlated. A competency profile is required for the performant ACTIVITY ABSTRACTIONS execution of professional/vocational ”activities” ONTOLOGY consists of BEHAVIORAL / SKILLS Each activity is abstracted using concepts and their SEMANTICS relations ©Synergetics 2012 All rights reserved
  • 27.
    Activity Semantics 1 Job Category • Competency framework (ROME CATALOG) 2 Associated competencies • Semantic Activity abstractions click 3 Semantic entities & relations 2 1 3
  • 28.
    Competency ANALOGY: similarities ActivitySemantics Compare similarities between 2 competencies • Job categories • Competencies belonging to a job category Score on 1 Common Relation ©Synergetics 2010 all rights resverved
  • 29.
    Competency GAP analysis: differencies Compare differences between 2 competencies Activity Semantics • Job categories • Competencies belonging to a job category Score on 1 Common Relation ©Synergetics 2010 all rights resverved
  • 30.
    Find candidates Specify vacancies 1. Select ranking threshold 2. Select (in)experienced candidates 3. Select the job offer to start the search ©Synergetics 2010 All rights reserved
  • 31.
    Find candidates All animals are equal! Click to visualise the competency Results Overview Semantic Ranking according to “presence” of relevant competencies
  • 32.
    Visualise the profile 1. Click the ‘bar icon’, multiple selections are allowed 3. Select graph type 2. Graphic representation of the “activity semantics” of the competency (indicators) ©Synergetics 2010 all rights resverved
  • 33.
    Find your candidate Visualise the semantic differences of the 7 competencies from the 9 selected candidates 7 competencies …but some are more equal than others  9 selected candidates ©Synergetics 2010 All rights reserved
  • 34.
    Conclusion • The “ActivitySemantics” extraction method makes use of Natural Language Technology : • Abstraction : Extract universal semantics from different competenvy descriptions • Interpretation : Skills and Competency extraction from employabilityPortfolios™, texts, CVs, vacancies and … Regional Employability Platform enabled systems! • For this demo we used (anonymised) real life data from 13.000 candidates and 13.000 vacancies. • The Activity Semantics method is based on its own ABAS Ontology • Semantic annotation & enrichment of existing ESCO and other European occupation, vocation, qualification and competency frameworks (ROME, COLO, NOS, UKCES, SSC, QCF, …etc) • This semantic method allows for considerable refined decision making when searching for the right candidate or vacancy, based upon a set of required or desired competencies. ©Synergetics 2010 All rights reserved
  • 35.
    Semantic DNA The Technology •Semantic metadata • Interpretation • Input: Textual data • Output: Semantic DNA • Comparison • Input: Semantic DNAs • Output: Scores of Similarity, Difference, Equivalence. These are the basis for semantic matching • Robust text understanding technology • Language parsing and interpretation • Customisable and optimisable • Languages (French, English currently, Dutch coming up)
  • 36.
    Use of CompetenceDNA • Operation • Extraction of competence semantics • Semantic comparison of competences • Customization by competence frameworks • Knowledge bases of competence frameworks • Language capability, based on knowledge bases • Comparison of competences: • free2ref • free2free
  • 37.
    Example free2free comparison Text Editor for Input Text Input Semantic DNA Semantic comparison