1. Ontology-based Access to Legacy Databases Oscar Corcho, Asunción Gómez-Pérez {asun,ocorcho}@fi.upm.es Ontological Engineering Group Laboratorio de Inteligencia Artificial Facultad de Informática Universidad Politécnica de Madrid Campus de Montegancedo sn, 28660 Boadilla del Monte, Madrid, Spain

2. –Migraci ó n Web Profunda  Web Semántica Uno de los principales impulsos para la Web Semántica sería la exposición en ella de la gran cantidad de bases de datos relacionales existentes en la Web, de modo que su contenido pueda ser procesado automáticamente. Tim Berners Lee 1998 Design Issues No tiene sentido tratar de anotar manualmente los millones de páginas existentes en la Web. Esas páginas están diseñadas para consumo humano. Para poblar la Web semántica hay que volverse hacia las bases de datos existentes y su contenido. Tim Berners Lee 2006 Entrevista iSIGHT –Explicitaci ó n de la sem á ntica de las bases de datos relacionales Los tres problemas de investigación más importantes en Bases de Datos solían ser ’rendimiento’, ’rendimiento’ y ’rendimiento’; en los próximos años los tres problemas más importantes y que verdaderamente supondrán un desafío para la investigación serán ’la semántica’, ’la semántica’ y ’la semántica’. Stefano Ceri 2004 SWDB Toronto –Reutilizar recursos no ontológicos y ontológicos para construir nuevas ontolog í as. G ó mez-P é rez, Su á rez-Figueroa, D5.3.1. NeOn deliverable Motivation

3. Different sources Motivation PublishGenerate Set of Standards

4. Hierarchy Models Motivation Different sources Modeled Generate

5. BDs Ontologías Heterogeneity problem Lenguaje Primitivas del paradigma Nombres del dominio Cobertura granularidad Posibles usos del modelo

6. Existing approaches 1.Construir una nueva ontología a partir de 1 esquema y datos de 1 BD (OntoStudio, KaOn Reverse) 2. Mapear la Onto construida en el enfoque 1, con una ontologia de legado (NeOn toolkit UKARL) 3. Mapear BD existente a una ontologia de legado (NeOn Toolkit UPM) a) Volcado masivo b) Dirigido por las consultas new ontology existing ontology 1 2 3

7. Existing approaches 1. Construir 1 Ontología a partir de 1 esquema y datos de 1 BD (OntoStudio, KaOn Reverse) http://kaon.semanticweb.org/alphaworld/reverse/

8. Existing Approaches 2. Mapear la onto construida en el enfoque 1, con una ontologia de legado (NeOn toolkit UKARL) http://www.neon-toolkit.org

9. Existing approaches 3.Acceso a contenido de BD usando ontologías de legado (NeOn toolkit at UPM) –R20 y ODEMapster Vocado Masivo No migrado masivo –Fundfinder Case study –FAO case study 4. Construcción de redes de ontologías mediante la reutilización y reingeniería de recursos no ontológicos y ontológicos –Seemp case study 3 4

10. Ontology-based view over a relational model (I) Punto Europeo PuntoGPS PuntoAsiatico = Aeropuertos f (Aeropuertos) Ontología O 2 Modelo Relacional M 1 PuntoEuropeo PuntoEspañol Estación Centro Comunicaciones Aeropuerto = f (Aeropuertos) Ontología O 1 RC (O 2,M 1 ) RC (O 1,M 1 )

11. BDR Modelo Relacional Personal Organización Pregunta: Nombre de los profesores de la universidad UPM * Un profesor es una persona cuyo puesto es “docente” * Una universidad es una organización de tipo “3” Consulta: valores de la columna nombre de los registros de la tabla Personal para los que el valor de la columna puesto is “docente” que estén relacionados con al menos un registro de la tabla Organización con el valor “3” en la columna tipo y “UPM” en la columna nombre. Descripción Formal de Correspondencia Procesado de la consulta de acuerdo a la descripción formal de correspondencia Ontología Profesor Doctorando Universidad Procesador Ontology-based view over a relational model (II)

12. Upgrading Database content to the semantic Web Integrating information from different DB sources Reuse of legacy DBs and legacy ontologies R 2 O: Declarative Mapping description language ODEMapster: Generic query processor. –asking queries to a relational database using ontology terms –On demand query answering –Batch ontology population A well defined method for upgrading and integrating content from heterogeneous sources.

13. Casos de mapping cubiertos por el lenguaje: R 2 O (Relational-to-Ontology) Language A view maps exactly one concept in the ontology. A subset of the columns in the view map a concept in the ontology. A subset (selection) of the records of a database view map a concept in the ontology. A subset of the records of a database view map a concept in the onto. but the selection cannot be made using SQL. A column in a database view maps directly an attribute or a relation. A column in a database view maps an attribute or a relation after some transformation. A set of columns in a database view map an attribute or a relation. One or more concepts can be extracted from a single data field (not in 1NF). para conceptos... para atributos...

14. ODEMapster: Volcado masivo Creación de un repositorio semántico en RDF

15. ODEMapster: Dirigido por las consultas

16. Lenguajes utilizados por ODEMapster

17. Fund Finder Case Study Esperonto Services IST-2001-34373 http://www.esperonto.net

18. The Overall Process How does it work?

19. Source system description Source system: FISUB Database (upgraded with R2O) –Maintained manually on a daily basis –Published on the Web (www.gencat.net) –Around 300 records, 5 tables (most of the information contained in one of them). –Search for fundings by: Dates Sector Subsector Keywords Purpose Source system: BOE Web site (Spain’s official journal) –PDF files published on the web (www.boe.es)www.boe.es –Search by date and number Source system: DOGC (Catalonia’s official journal) –HTML files published on the web (www.gencat.net/diari)www.gencat.net/diari –Search by date and number

20. Build the ontologies –Funding opportunity ontology: Concepts: 32 Instance attributes: 12 Subclass-of: 25 Ad-hoc relations: 8 –Funding Body ontology: Concepts: 7 Subclass-of: 5 Ad-hoc relations: 2 –Applicant ontology: Concepts: 22 Instance attributes: 4 Subclass-of: 18 Ad-hoc relations: 4 –Official Publication ontology: Concepts: 9 Instance attributes: 6 Subclass-of: 7 Ad-hoc relations: 1 –Organization ontology: Concepts: 6 Instance attributes: 10 Ad-hoc relations: 6 –Person ontology: Concepts: 5 Instance attributes: 9 Ad-hoc relations: 4 –Location ontology: Concepts: 4 Instance attributes: 2 Subclass-of: 3 Ad-hoc relations: 3 Inter ontology relations. WHAT? WHO? FOR WHOM? WHERE? PUBLISHED?

21. Population example (I) Attibute Direct Mapping How to express this using R 2 0?? Attibute Mapping with transformation (Regular Expression) Relation Mapping w. Transformation (Regular Expression) Relation Mapping w. Transformation (Keyword search)

22. Population example (II) The Operation element defines a transformation based on a regular expression to be applied to the database column for extracting property values Population example (II)

23. Population example (III) ConceptMap-Def element: Describes how instances of a concept are extracted from the database. AttributeMap-Def element: Describes how values for an attribute are extracted from the database.

24. The Overall Process How does it work?

26. The Overall Process How does it work? Content from different sources is aggregated

27. (Conceptual model) Taxonomy Ontologies Instances Implementation List of Instances Visualization

28. FAO Scenario The ontologies produced with this framework will be used by the Food and Agriculture Organization of the United Nations (FAO) in many different large applications such the Fisheries Stock Depletion Assessment System. Slide 28

29. Fisheries Ontologies Lifecycle Slide 29

30. 2. Ontology Population Slide 30 R20 and X2O are being used in the Ontology population activity

31. FAO Case study Land areas Fishing areas Biological entities Fisheries commodities Vessel types and size Gear types R 2 O Document R 2 O Document R 2 O Document R 2 O Document R 2 O Document R 2 O Document FAO FIGIS DB http://www.fao.org/aims/aos/fi/

32. DB 1 FAO Ontologies Query Driven (on demand process) DB 2DB 3 ODEMaspter R 2 O Document Client

33. RDF(S) OWL Ontology Development O. Specification O. Conceptualization O. Implementation O. Formalization 1 Thesauries DB Lexicons Resources Knowledge Resource Reuse Study Knowledge Resource Reengineering 2 3 Ontology Reuse Study: (Searching, Evaluation/Assesment, Selection) O. Repositories Flogic RDF(S) OWL Ontology Reestruturation (Pruning, Extension, Specialization, Modularization) 8 O. Localization 9 Maintain Use O. Assessment 10 Ontology Support Activities: Elicitation; Documentation; Configuration Management; Evaluation (V&V) 1,2,3,4,5,6,7,8,9 Ontology Reengineering: ( Reverse Engineering, Reestructuration Forward Engineering) 4 6 7 Scenarios in the ontology life cycle Flogic O. Alignment O. Merge 5

34. Resources O. Repositories and Registries Flogic RDF(S) OWL Thesauri DB Lexicons Glossaries O. Specification O. Conceptualization O. Implementation O. Formalization 1 Non Ontological Resource Reuse Non Ontological Resource Reengineering 2 3 Ontology Reuse Ontology Restructuring (Pruning, Extension, Specialization, Modularization) 7 O. Localization 8 Ontology Support Activities: Knowledge Acquisition (Elicitation); Documentation; Configuration Management; Evaluation (V&V); Assessment 1,2,3,4,5,6,7,8 O. Aligning O. Merging Aligments Ontology Reengineering 4 6 5 5 RDF(S) OWL Flogic 46 2 2 3 4 5 6 6 Scenarios in the ontology life cycle

35. Existing approaches 4. Construcción de redes de ontologías mediante la reutilización y reingeniería de recursos no ontológicos y ontológicos –Seemp case study 4

36. g ES Lombard ES (It) q ES r ES p ES a ES c ES i ES n ES e ES h ES l ES o ES f ES d ES m ES b ES Requester ES Responding ES ES not involved Job Seeker’s Candidacy Employer Job Vacancy LEGENDA Helping Job Seekers on their way EuropeanEmploymentMediatorsMarketplace Local Matching algorithm EURES ES (Int) Local Matching algorithm Private ES (Int) Local Matching algorithm Wallonia ES (Be) Local Matching algorithm Catalonia ES (Es)

37. Ms Centralized network of ontologies 1. Build a reference ontology Federated network of ontologies 1.Build a reference ontology for the domain 2.Build local ontologies 3.Build mappings between the core and local ontologies 4.Build mappings between the local ontologies and the data sources Ms 2. Build mappings between the reference ontology and the data sources

38. ISCO-88 (COM), ONET, EURES taxonomy, FOET, ISCED97, NACE, ISO 4217, ISO 3166, ISO 6392, HR-XML, … RDF(S) OWL Repositories & libraries ES Data Sources BuildingReferenceOntology Reference Ontology BuildingLocalOntologies Local Ontologies BuildingMappings L.O. - ES Data Sources Mappings L.O. - ES Data Sources BuildingMappings R.O. - L.O. Mappings

39. SEEMP Ontology Network Life Cycle: Iterative model life cycle O. Elicitation O. Documentation O. LocalizationO. Pruning O. Extension O. Specialization O. Specification O. Conceptualization O. Evaluation O. Implementation RDF(S) OWL Maintain Use O. Formalization O. Assessment O. Searching O. Selection RDF(S) OWL Repositories & libraries ISCO-88 (COM), ONET, EURES taxonomy, FOET, ISCED97, NACE, ISO 4217, ISO 3166, ISO 6392, HR-XML, … Assessment Select Search Reengineer

40. O. Elicitation O. Documentation O. LocalizationO. Pruning O. Extension O. Specialization O. Specification O. Conceptualization O. Evaluation O. Implementation RDF(S) OWL Maintain Use O. Formalization O. Assessment Assessment Select Search O. Assessment O. Searching O. Selection RDF(S) OWL Repositories & libraries ISCO-88 (COM), ONET, EURES taxonomy, FOET, ISCED97, NACE, ISO 4217, ISO 3166, ISO 6392, HR-XML, … Ontology Specification Reengineer

41. Ontology Specification 60 Competency questions grouped into 5 categories (modular approach) –Job Seeker (12) What is his/her education level? –Job Offer (12) What are the required skills for the job offer? –Time and date management (7) When the job seeker completed his/her first degree? –Currencies (4) The offered salary is given in US dollars? –General (25) Given the employer information, economic activity of the employer and the job offer profile (job, contract type, salary, work condition, contract duration), what job seekers are the most appropriate? Given the job offer profile (job, contract type, salary, work condition) and the required profile to seek (required education level, required work experience, required knowledge, required skills), what job seekers are the most appropriate? Classes: Contract Type, Compensation, Work Condition, Job Seeker, Job Offer … Relations: has job category, has compensation, requires work experience … Attributes: Name, date of birth, email … Each organization has job offers for job seekers Vocabulary: Questions: contract type, salary, work condition, job seeker, job offer, … Answers: autonomous, 3000 euro, holliday job, …

42. O. Elicitation O. Documentation O. LocalizationO. Pruning O. Extension O. Specialization O. Specification O. Conceptualization O. Evaluation O. Implementation RDF(S) OWL Maintain Use O. Formalization O. Assessment O. Searching O. Selection RDF(S) OWL Repositories & libraries ISCO-88 (COM), ONET, EURES taxonomy, FOET, ISCED97, NACE, ISO 4217, ISO 3166, ISO 6392, HR-XML, … Knowledge reuse resource study Assessment Select Search Reengineer

43. Search and Assess Standards and Taxonomies We select the most appropriate standards and taxonomies for: –Occupation Classification ISCO-88 (COM), SOC, ISCO-88, ONET, Eures Taxonomy. –Classification of Economic Activities ISIC Rev. 3.1, NACE Rev. 1.1, NAICS –Apprenticeship classifications ISCED 97, FOET –Currency Classification ISO 4217 –Geography Classification ISO 3166, Eures Taxonomy Language Classification ISO 6392, CEF Driving License Classification European Legislation Skill Classification Eures Taxonomy Contract Types Classification LE FOREM, Eures and BLL Classification Work Condition Classification LE FOREM, Eures and BLL Classification Assessment activity : Matching terminology from Competency Questions against the Standards Assessment Select Search Reengineer

44. Selection of Standards The current European needs The degree of coverage FOETISCED 97NAICSNACE Rev. 1.1 ISIC Rev. 3.1 ISCO-88 (COM) ISCO-88ONETSOC Apprenticeship Classification Classification of Economic Activities Occupation Classification Reference Ontology shall be based on the international, European or de-facto industrial standards But, we need also proprietary taxonomies … Currency Classification Geography Classification Language Classification Driving License ISO 4217ISO 3166ISO 6392Community Driving License Assessment Select Search Reengineer

45. Reengineering resources Oracle DB HTML MS Access ISCO-88 (COM) EURES Taxonomy (proprietary) ONET Prune Integrate OccupationOntology Ad hoc wrapper WSML exporter Extend Specialize Assessment Select Search Reengineer

46. Reengineering resources ISO 3166-1 (XML) Regions Table (Eures Oracle DB) …. SPAIN ES … Location CountryRegion subClass-Of has region SpainCataluña Canarias Galicia Andalucía Ontology model Ontology instances Excerpt of the Geography Ontology

47. O. Elicitation O. Documentation O. LocalizationO. Pruning O. Extension O. Specialization O. Specification O. Conceptualization O. Evaluation O. Implementation RDF(S) OWL Maintain Use O. Formalization O. Assessment O. Searching O. Selection RDF(S) OWL Repositories & libraries ISCO-88 (COM), ONET, EURES taxonomy, FOET, ISCED97, NACE, ISO 4217, ISO 3166, ISO 6392, HR-XML, … Ontology reuse Assessment Select Search Reengineer

48. Assessing Time Ontologies 1. Identification of criteria for comparing the candidate set of temporal ontologies 2. Assess all existing temporal ontologies against the criteria Different temporal granularities Concatenation of intervals Explicit modeling of proper intervals Distinction between open and closed intervals Convex and non convex intervals Relations between time intervals Absolute and Relative Time Time Interval Time Points Cyc’s Upper Ontology Unrestricted Time Ontology Simple Time Ontology Reusable Time Ontology Kestrel Time Ontology SRI’s Time Ontolog y SUMO Time Ontology DAML Time Ontology AKT Time Ontology Time Points Time Interval Absolute and Relative Time Relations between time intervals Convex and non convex intervals Distinction between open and closed intervals Explicit modeling of proper intervals Concatenation of intervals Different temporal granularities Provides axioms O. Assessment O. Searching O. Selection

49. Process for assessing Time Ontologies (II) 3.Checking which temporal properties are needed for answering the Competency questions a.When the job seeker completed his/her first degree? b.Is the job seeker older than 30 years? c.How much time did the job seeker spend completing his/her first degree? d.How long is the duration of the contract? e.Which job offers were posted in last 24 hours? f.Which job offers were posted in last 7 days? ? g.Which job offers were posted in last month? h.Was the job seeker unemployed? i.Was the job seeker a student between 1995 and 2000? 4.Checking which temporal properties are needed for answering the Competency questions Different temporal granularities Concatenation of intervals Explicit modeling of proper intervals Distinction between open and closed intervals Convex and non convex intervals Relations between time intervals Absolute and Relative Time Time Interval Time Points b, c a,d,f,g a h i O. Searching O. Assessment O. Selection

50. The Time Ontology Selection Cyc’s Upper Ontology Unrestricted Time Ontology Simple Time Ontology Reusable Time Ontology Kestrel Time Ontology SRI’s Time Ontolog y SUMO Time Ontology DAML Time Ontology AKT Time Ontology Time Points Time Interval Absolute and Relative Time Relations between time intervals Convex and non convex intervals Distinction between open and closed intervals Explicit modeling of proper intervals Concatenation of intervals Different temporal granularities Provides axioms O. Searching O. Assessment O. Selection

51. Conceptualization O. Elicitation O. Documentation O. LocalizationO. Pruning O. Extension O. Specialization O. Specification O. Conceptualization O. Evaluation O. Implementation RDF(S) OWL Maintain Use O. Formalization O. Assessment O. Searching O. Selection RDF(S) OWL Repositories & libraries ISCO-88 (COM), ONET, EURES taxonomy, FOET, ISCED97, NACE, ISO 4217, ISO 3166, ISO 6392, HR-XML, … Assessment Select Search Reengineer

52. Conceptualization: Modular approach for ontology construction Representation Ontology: WSML General/Common Ontologies: Time, Geography, Language Domain O.: Economic Activity, Occupation, Education, Skill, Driving License, Compensation, Labour Regulatory, Competence Application Domain O. : Job Seeker, Job Offer - + Reusability - + Usability

53. Reference Ontology Labour Regulatory Ontology Skill Ontology Language Ontology Occupation Ontology Geography Ontology Time Ontology Education Ontology Driving License Ontology Compensation Ontology Economic Activity Ontology Job Offer Ontology Job Seeker Ontology has work condition / is associated with has contract type / is associated with is located in / has salary / is associated with requires education / is associated with has activity sector / is associated with has nationality from / is nation of resides in / is residence of has salary / has contract type / is associated to has work condition / is associated to has location / is associated with has activity sector / is associated with has activity sector / is associated with has job category / is associated with has job category / Is associated with has education / is education of has mother tongue / is mother tongue of speaks / is spoken by has language proficiency / belongs to LE FOREM + BLL + EURES EURES ISO 6392 CEF ISCO-88 COM ONET EURES ISO 3166 EURES DAML Time Ontology FOET ISCED97 NACE Rev. 1.1 European Legislation ISO 4217 Ad hoc wrapper External Sources is associated with has job category / is associated to has date of birth / is date of birth of has begin date / is begin date of Competence Ontology subClass-Of requires competence / is associated with has competence / is competence of

54. Candidacy Offered Work Experience Objective ICT Objective Job Seeker Vacancy Organization Requested Work Experience ICT Vacancy Job Vacancy Competence Education Language Contract Type Compensation Work Condition Occupation Sector Location Country Computing Professionals has candidacy/ belongs to has objective / belongs to subClass-Of has job category is associated with subClass-Of has nationality from / is nation of resides in / is residence of has mother language / is mother tongue of speaks / is spoken by has competence / is competence of has education / is education of has work experience / belongs to has work condition / is associated to has contract type / is associated to has compensation / is associated to is associated with / has location has activity sector / is associated with has activity sector / is associated with has job category / is associated with has job category / is associated with has job vacancy/ belongs to has location / is location of has vacancy/ belongs to subClass-Of has job category / is associated with is associated with / requires work experience has activity sector / is associated with has job category/ is associated with requires education / is associated with requires competence / is associated with has work condition / is associated with has contract type / is associated with has compensation / is associated with is located in / is associated with has job category/ is associated with has activity sector / is associated with Job Offer Ontology Job Seeker Ontology OccupationOntology LanguageOntology EducationOntology CompetenceOntology LabourRegulatoryOntology CompensationOntology GeographyOntology EconomicActivityOntology Details of the ontology

55. ISCO-88 (COM), ONET, EURES taxonomy, FOET, ISCED97, NACE, ISO 4217, ISO 3166, ISO 6392, HR-XML, … RDF(S) OWL Repositories & libraries ES Data Sources BuildingReferenceOntology Reference Ontology BuildingLocalOntologies

56. Local Ontologies Building Process Option 1: Building Local Ontologies from the Reference Ontology. Reference Ontology Specialize Extend Resultant Local Ontology Prune Option 2: Building Local Ontologies as a reverse engineering process from ES Data Sources. ES Data Sources Reverse Engineering Resultant Local Ontology

57. Hybrid approach for building Local Ontologies A hybrid approach Reference Ontology Job Offer Job Seeker Ontology ReverseEngineering Skill Education Economic Activity Occupation Ontology Local Ontology ES Occupation Taxonomy Integrate Option 1 for Job Seeker and Job Offer Ontologies Option 1 for Job Seeker and Job Offer Ontologies Option 2 for Occupation, Education, etc. Option 2 for Occupation, Education, etc.

58. ISCO-88 (COM), ONET, EURES taxonomy, FOET, ISCED97, NACE, ISO 4217, ISO 3166, ISO 6392, HR-XML, … RDF(S) OWL Repositories & libraries ES Data Sources BuildingReferenceOntology Reference Ontology BuildingLocalOntologies Local Ontologies BuildingMappings R.O. - L.O. Mappings BuildingMappings L.O. - ES Data Sources XMapsterMappings L.O. - ES Data Sources X2OX2O

59. Conclusions Mappings Onto-BD –Estudio y caracterización de las diferentes situaciones en el establecimiento de correspondencias entre ontologías y BD –Propuesta de un modelo para la definición declarativa de dichas correspondencias –Implementación de dicho modelo en un lenguaje formal, declarativo y suficientemente expresivo para describir situaciones complejas : R 2 O –Definición de un procesador capaz de llevar a cabo la traducción de consultas entre modelos: Procesador ODEMapster & Lenguaje ODEMQL –Modos de ejecución: Dirigido por las consultas y migración masiva

60. Acknowledgement Mari Carmen Suárez-Figueroa Jesús Barrasa Boris Villazón Raúl Palma NeOn project partners SEEMP project partners