Towards an industry-wide
ifcOWL: choices and issues
Pieter Pauwels, Ghent University
María Poveda Villalón, UPM
3rd workshop on Linked Data in Architecture and Construction – 17 July 2015 - Eindhoven
• Diverse suggestions in academic research to make IFC available as an OWL
ontology to allow the usage of semantic web technologies
 Schevers and Drogemuller, 2005
 Beetz et al., 2009
• General purpose initiative to convert EXPRESS schemas and STEP file to OWL
ontology, not focused on IFC
 Krima et al., 2009 (OntoSTEP)
 Barbau et al., 2012 (OntoSTEP)
• From 2012 onwards, increasing uptake in research and developments
=> many different ifcOWL structures floating around
• Need for formalisation and standardisation
=> combine efforts and proceed in a more formalised, standard fashion
=> role of standardisation bodies: W3C, BuildingSMART
• BuildingSMART Technical Summit March 2015 (London) and October 2015
(Singapore)
we owe our past and present - 2005-2015
1. Short intro to ifcOWL Pieter and Walter +
comparison with Nam’s work
2. Scope: the extended ifcOWL
3. Deciding on ifcOWL open issues
FOCUS for this
presentation
outline
SHORT INTRO TO IFCOWL (W&P)
Image courtesy: Nam Vu Hoang, Aalto University
conversion procedure / approach Nam
DOWNSTREAM
Image courtesy: Nam Vu Hoang, Aalto University
conversion procedure / approach Nam
DOWNSTREAM
EXTENDED IFCOWL
Image courtesy: Nam Vu Hoang, Aalto University
conversion procedure / approach Nam
DOWNSTREAM
EXTENDED IFCOWL
FILE-BASED CONVERSION RELYING
ON THE IFCOWL ONTOLOGY
Image courtesy: Nam Vu Hoang, Aalto University
conversion procedure / approach Nam
conversion procedure EXPRESS schema to OWL
IFC
Schema
Simple data type
Defined data type
Aggregation data type
SET data type --------
LIST & ARRAY data type --------
Constructed data type
SELECT data type --------
ENUMERATION data type --------
Entity data type
Attributes --------
Derive attr
WHERE rules
Functions
Rules
ifcOWL
Ontology
owl:class + owl:DatatypeProperty restriction
owl:class
owl:class
-------- owl:ObjectProperty restriction on
ifc:hasSet
-------- indirect subclass of ifc:List
owl:class
-------- owl:unionOf ( owl:classes )
-------- one of ( owl:NamedIndividuals )
owl:class
-------- object properties
-
-
-
-
what is included in ifcOWL extended?
Image courtesy: Nam Vu Hoang, Aalto University
EXTENDED IFCOWL
Totally okay with ‘compartmentalization’ or ‘layers’
If it helps industry
differences (1)
Image courtesy: Nam Vu Hoang, Aalto University
differences (2)
Image courtesy: Nam Vu Hoang, Aalto University
OWL DL focus only
==> everything in there, should be enough
differences (3)
Image courtesy: Nam Vu Hoang, Aalto University
Leads to property renaming
==> 1 domain / range for each property
differences (4)
Image courtesy: Nam Vu Hoang, Aalto University
Wrap XSD types
differences (5)
Image courtesy: Nam Vu Hoang, Aalto University
Subclasses of simple datatypes
or other defined datatypes
differences (6)
Image courtesy: Nam Vu Hoang, Aalto University
| Can easily be changed
|
|
|
|
|
|
|
Step 2. Simple data types
ifc:REAL
rdf:type owl:Class ;
rdfs:subClassOf
[
rdf:type owl:Restriction ;
owl:allValuesFrom xsd:double ;
owl:onProperty ifc:has_double
] .
ifc:has_double
rdf:type owl:DatatypeProperty ;
rdf:type owl:FunctionalProperty ;
rdfs:label "has_double" ;
rdfs:domain
[
rdf:type owl:Class ;
owl:unionOf ( ifc:REAL )
] ;
rdfs:range xsd:double .
OWL
Step 3. Defined data types
IFC
Schema
Simple data type
Defined data type
ifcOWL
Ontology
owl:class + owl:DatatypeProperty restriction
owl:class
TYPE IfcAreaDensityMeasure = REAL;
END_TYPE;
ifc:IfcAreaDensityMeasure
rdf:type owl:Class ;
rdfs:subClassOf ifc:REAL .
EXPRESS
OWL
differences (6)
Image courtesy: Nam Vu Hoang, Aalto University
differences (7)
Image courtesy: Nam Vu Hoang, Aalto University
MANY OPTIONS HERE – LARGE IMPACT, as all geometric data (90% of IFC) is in
lists ==> huge impact on triple count
differences (8)
Image courtesy: Nam Vu Hoang, Aalto University
No policy here => we focused only on EXPRESS to OWL
RDF instances is a whole alternative area. If the Extended OWL Ontology is
followed, one should have enough and naming should not matter too much.
SCOPE: THE EXTENDED IFCOWL
VERSION
Image courtesy: Nam Vu Hoang, Aalto University
current status
ifcOWL
Walter
ifcOWL
Pieter
= identic. =
~ equiv. ~
Image courtesy: Nam Vu Hoang, Aalto University
targeted status
ifcOWL - e
Walter
ifcOWL - e
Pieter
= identic. =
= identic. =
ifcOWL - si
Walter
ifcOWL - si
Pieter
ifcOWL - st
Walter
ifcOWL - st
Pieter
= identic. =
= identic. =
= identic. =
= identic. =
importimport
Minimal aim for LDAC2015
DECIDING ON IFCOWL OPEN ISSUES
issue 1 - ontology name and provenance
issue 2 - URI naming convention
for object properties
issue 2 - URI naming convention
for object properties
Name_of_IfcRootName_of_IfcOrganization …Degree
issue 2 – alternative 1
issue 2 – alternative 1
Name
Name_of_IfcRootName_of_IfcOrganization …Degree
issue 2 – alternative 2
issue 2 – alternative 2
Name
Name_of_IfcRootName_of_IfcOrganization …Degree_of_IfcBSplineCurve
Degree
STANDARD / SIMPLE
EXTENDED (FULL)
issue 2 – alternative 3
issue 2 – alternative 3
Name_of_IfcRootName_of_IfcOrganization …Degree_of_IfcBSplineCurve
issue 3 - URI naming convention
for individuals in Enumerations
=> Similar to object property renaming, except for the domain and range restrictions
=> Alternatives: use all short names (ENUM individuals belong to multiple classes) or
use all long names ([]_of_[]) (ENUM individuals belong to one class only)
RECAP / CONCLUSIONS
Parallels in the extended ifcOWL (1)
Image courtesy: Nam Vu Hoang, Aalto University
OWL DL
Wrap XSD types
Subclasses of simple datatypes
or other defined datatypes
|
|
|
|
|
|
|
OWL DL
1
Can we follow the subproperty proposal?
Parallels in the extended ifcOWL (2)
Image courtesy: Nam Vu Hoang, Aalto University
MANY OPTIONS HERE – LARGE IMPACT, as all geometric data (90% of IFC) is in
lists ==> huge impact on triple count
Beyond development of an extended ifcOWL
Image courtesy: Nam Vu Hoang, Aalto University
explore options for compartmentalisation
ifcOWL - e
Walter
ifcOWL - e
Pieter
= identic. =
= identic. =
ifcOWL - si
Walter
ifcOWL - si
Pieter
ifcOWL - st
Walter
ifcOWL - st
Pieter
= identic. =
= identic. =
= identic. =
= identic. =
importimport
Minimal aim for LDAC2015
Thank you
Pieter Pauwels, Ghent University
pipauwel.pauwels@ugent.be
María Poveda Villalón, UPM Madrid
mpoveda@fi.upm.es

LDAC 2015 - Towards an industry-wide ifcOWL: choices and issues

  • 1.
    Towards an industry-wide ifcOWL:choices and issues Pieter Pauwels, Ghent University María Poveda Villalón, UPM 3rd workshop on Linked Data in Architecture and Construction – 17 July 2015 - Eindhoven
  • 2.
    • Diverse suggestionsin academic research to make IFC available as an OWL ontology to allow the usage of semantic web technologies  Schevers and Drogemuller, 2005  Beetz et al., 2009 • General purpose initiative to convert EXPRESS schemas and STEP file to OWL ontology, not focused on IFC  Krima et al., 2009 (OntoSTEP)  Barbau et al., 2012 (OntoSTEP) • From 2012 onwards, increasing uptake in research and developments => many different ifcOWL structures floating around • Need for formalisation and standardisation => combine efforts and proceed in a more formalised, standard fashion => role of standardisation bodies: W3C, BuildingSMART • BuildingSMART Technical Summit March 2015 (London) and October 2015 (Singapore) we owe our past and present - 2005-2015
  • 3.
    1. Short introto ifcOWL Pieter and Walter + comparison with Nam’s work 2. Scope: the extended ifcOWL 3. Deciding on ifcOWL open issues FOCUS for this presentation outline
  • 4.
    SHORT INTRO TOIFCOWL (W&P)
  • 5.
    Image courtesy: NamVu Hoang, Aalto University conversion procedure / approach Nam
  • 6.
    DOWNSTREAM Image courtesy: NamVu Hoang, Aalto University conversion procedure / approach Nam
  • 7.
    DOWNSTREAM EXTENDED IFCOWL Image courtesy:Nam Vu Hoang, Aalto University conversion procedure / approach Nam
  • 8.
    DOWNSTREAM EXTENDED IFCOWL FILE-BASED CONVERSIONRELYING ON THE IFCOWL ONTOLOGY Image courtesy: Nam Vu Hoang, Aalto University conversion procedure / approach Nam
  • 9.
    conversion procedure EXPRESSschema to OWL IFC Schema Simple data type Defined data type Aggregation data type SET data type -------- LIST & ARRAY data type -------- Constructed data type SELECT data type -------- ENUMERATION data type -------- Entity data type Attributes -------- Derive attr WHERE rules Functions Rules ifcOWL Ontology owl:class + owl:DatatypeProperty restriction owl:class owl:class -------- owl:ObjectProperty restriction on ifc:hasSet -------- indirect subclass of ifc:List owl:class -------- owl:unionOf ( owl:classes ) -------- one of ( owl:NamedIndividuals ) owl:class -------- object properties - - - -
  • 10.
    what is includedin ifcOWL extended? Image courtesy: Nam Vu Hoang, Aalto University EXTENDED IFCOWL Totally okay with ‘compartmentalization’ or ‘layers’ If it helps industry
  • 11.
    differences (1) Image courtesy:Nam Vu Hoang, Aalto University
  • 12.
    differences (2) Image courtesy:Nam Vu Hoang, Aalto University OWL DL focus only ==> everything in there, should be enough
  • 13.
    differences (3) Image courtesy:Nam Vu Hoang, Aalto University Leads to property renaming ==> 1 domain / range for each property
  • 14.
    differences (4) Image courtesy:Nam Vu Hoang, Aalto University Wrap XSD types
  • 15.
    differences (5) Image courtesy:Nam Vu Hoang, Aalto University Subclasses of simple datatypes or other defined datatypes
  • 16.
    differences (6) Image courtesy:Nam Vu Hoang, Aalto University | Can easily be changed | | | | | | |
  • 17.
    Step 2. Simpledata types ifc:REAL rdf:type owl:Class ; rdfs:subClassOf [ rdf:type owl:Restriction ; owl:allValuesFrom xsd:double ; owl:onProperty ifc:has_double ] . ifc:has_double rdf:type owl:DatatypeProperty ; rdf:type owl:FunctionalProperty ; rdfs:label "has_double" ; rdfs:domain [ rdf:type owl:Class ; owl:unionOf ( ifc:REAL ) ] ; rdfs:range xsd:double . OWL
  • 18.
    Step 3. Defineddata types IFC Schema Simple data type Defined data type ifcOWL Ontology owl:class + owl:DatatypeProperty restriction owl:class TYPE IfcAreaDensityMeasure = REAL; END_TYPE; ifc:IfcAreaDensityMeasure rdf:type owl:Class ; rdfs:subClassOf ifc:REAL . EXPRESS OWL
  • 19.
    differences (6) Image courtesy:Nam Vu Hoang, Aalto University
  • 20.
    differences (7) Image courtesy:Nam Vu Hoang, Aalto University MANY OPTIONS HERE – LARGE IMPACT, as all geometric data (90% of IFC) is in lists ==> huge impact on triple count
  • 21.
    differences (8) Image courtesy:Nam Vu Hoang, Aalto University No policy here => we focused only on EXPRESS to OWL RDF instances is a whole alternative area. If the Extended OWL Ontology is followed, one should have enough and naming should not matter too much.
  • 22.
    SCOPE: THE EXTENDEDIFCOWL VERSION
  • 23.
    Image courtesy: NamVu Hoang, Aalto University current status ifcOWL Walter ifcOWL Pieter = identic. = ~ equiv. ~
  • 24.
    Image courtesy: NamVu Hoang, Aalto University targeted status ifcOWL - e Walter ifcOWL - e Pieter = identic. = = identic. = ifcOWL - si Walter ifcOWL - si Pieter ifcOWL - st Walter ifcOWL - st Pieter = identic. = = identic. = = identic. = = identic. = importimport Minimal aim for LDAC2015
  • 25.
    DECIDING ON IFCOWLOPEN ISSUES
  • 26.
    issue 1 -ontology name and provenance
  • 27.
    issue 2 -URI naming convention for object properties
  • 28.
    issue 2 -URI naming convention for object properties Name_of_IfcRootName_of_IfcOrganization …Degree
  • 29.
    issue 2 –alternative 1
  • 30.
    issue 2 –alternative 1 Name Name_of_IfcRootName_of_IfcOrganization …Degree
  • 31.
    issue 2 –alternative 2
  • 32.
    issue 2 –alternative 2 Name Name_of_IfcRootName_of_IfcOrganization …Degree_of_IfcBSplineCurve Degree STANDARD / SIMPLE EXTENDED (FULL)
  • 33.
    issue 2 –alternative 3
  • 34.
    issue 2 –alternative 3 Name_of_IfcRootName_of_IfcOrganization …Degree_of_IfcBSplineCurve
  • 35.
    issue 3 -URI naming convention for individuals in Enumerations => Similar to object property renaming, except for the domain and range restrictions => Alternatives: use all short names (ENUM individuals belong to multiple classes) or use all long names ([]_of_[]) (ENUM individuals belong to one class only)
  • 36.
  • 37.
    Parallels in theextended ifcOWL (1) Image courtesy: Nam Vu Hoang, Aalto University OWL DL Wrap XSD types Subclasses of simple datatypes or other defined datatypes | | | | | | | OWL DL 1 Can we follow the subproperty proposal?
  • 38.
    Parallels in theextended ifcOWL (2) Image courtesy: Nam Vu Hoang, Aalto University MANY OPTIONS HERE – LARGE IMPACT, as all geometric data (90% of IFC) is in lists ==> huge impact on triple count Beyond development of an extended ifcOWL
  • 39.
    Image courtesy: NamVu Hoang, Aalto University explore options for compartmentalisation ifcOWL - e Walter ifcOWL - e Pieter = identic. = = identic. = ifcOWL - si Walter ifcOWL - si Pieter ifcOWL - st Walter ifcOWL - st Pieter = identic. = = identic. = = identic. = = identic. = importimport Minimal aim for LDAC2015
  • 40.
    Thank you Pieter Pauwels,Ghent University pipauwel.pauwels@ugent.be María Poveda Villalón, UPM Madrid mpoveda@fi.upm.es