Post on 12-Apr-2017
transcript
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
IfcWoD (Web of Data)Semantically Adapting IFC Model Relations
into OWL Properties
Tarcisio Mendes de Farias, Ana Roxin andChristophe Nicolle
Standards Summit – RotterdamApril 11 - April 14
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Agenda
Motivation
• General context presentation
• Semantic Web technologies for IFC
Approach
• General presentation of ifcWOD
• Semantic adaptation of IFC relations
Benefits
• Simplifying query writing
• Reducing query execution time
• Reducing data redundancy
April 12th 2016 ifcWOD 2
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
MOTIVATION
April 12th 2016 ifcWOD 3
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
General Context
Linked data
• Good practices for publishing and connecting data on the Semantic Web
• URIs
• Network Protocol HTTP
• Semantic Web standards RDF
• Links among vocabularies
Building Information Modeling (BIM)
• IFC (Industrial Foundation Classes) is the ISO standard for BIM
• Information exchange based on STEP file format
How Linked Data (LD) can help in the context of BIM ?
• Enhancing BIM interoperability
• Static format (file-based) vs RDF
April 12th 2016 ifcWOD 4
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Semantic Web technologies for IFC
OWL for IFC
EXPRESS vs. OWL
• STEP-based files vs RDF triples
• Object-oriented vs graph modelling
Semantically adapting IFC into OWL
• IFC relationships
Intuitive building information manipulation
ifcOWL
IFC-based ontology
• Direct syntax mapping of IFC EXPRESS specification
BuildingSMART Linked Data Working Group
W3C Linked Building Data Community Group
April 12th 2016 ifcWOD 5
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
ifcOWL drawbacks
Understanding of IFC object properties and relationships
Access to the semantics of building data
Correct application of Linked Data principles
No leverage from constraints dictated by the EXPRESS/STEP file format
April 12th 2016 ifcWOD 6
“An ontology is a formal, explicit specification of a shared conceptualization”
Studer, R., Benjamins, R. & Fensel, D., 1998. Knowledge engineering: Principles and methods. Data & Knowledge Engineering, 25(1–2), p. 161–198.
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
PROPOSED APPROACH
April 12th 2016 ifcWOD 7
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
ifcWoD, IFC Web of Data Ontology
◼ Semi-automatically conceived, based on: IFC EXPRESS model
IfcOWL
Property Set Definition (PSD)
◼ Main contributions:
April 12th 2016 ifcWOD 8
Adaptation of existing IFC properties
ifcRelationship
ifcProperty
New properties
ifcwod:isDefinedBy_ifcObject
ifcwod:operatesOn
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Adapting ifcRelationship into OWL
◼ We semantically adapt ifcRelationship entities into OWL
Inverse and Relating[name of relating object] attributes
April 12th 2016 ifcWOD 9
Our approach can be applied on all 41 ifcRelationship subtypes of the IFC standard.
Only 14 of these subtypes have additional semantic attributes, for which the ifcOWL representation is needed.
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Adaptation of ifcRelationship
◼ IFC properties and relationships in the IfcWOD ontology schema
April 12th 2016 ifcWOD 10
ifcowl:IfcRelDefinesByProperties
ifcowl:IfcObjectDefinition ifcowl:IfcPropertySetDefinitionSelect
ifcowl:IfcWallStandardCase ifcowl:IfcPropertySet
rdfs:subClassOf rdfs:subClassOf
ifcowl:RelatingPropertyDefinition_of_IfcRelDefinesbyProperties
ifcowl:RelatedObjects_of_IfcRelDefinesbyProperties
ifcwod:isDefinedBy_IfcObject
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Adaptation of ifcProperty
◼ Idea: take advantage of PSDs
◼ ifcWoD: an IFC property is part of the schema
Structured data
Mapping rules from existing XSD files (Pset)
April 12th 2016 ifcWOD 11
Mapping
Rule
Property Set Definition
(PSD) XSD
OWL
language
R1 <xs:complexType name="PropertyDef"> owl:ObjectProperty
R2 <xs:element type="xs:string" name="Name"> rdfs:label and URN
R3<xs:element type="xs:string"
name="Definition">rdfs:comment
R4 <xs:element name="NameAliases"> rdfs:label (@lang)
R5 <xs:element name="DefinitionAliases"> rdfs:comment (@lang)
R6<xs:element type="PropertyType"
name="PropertyType">
rdfs:subPropertyOf and
rdfs:range
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
BENEFITS
April 12th 2016 ifcWOD 12
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Main advantages
IfcWoD allows enhancing reasoning
Easier query writing and relationship understanding
Reduction of query execution time
Reduction of data redundancy
April 12th 2016 ifcWOD 13
OWL transitive properties e.g.: ifcwod:isPredecessorTo_IfcProcess
SELECT ?x ?z { ?x ifcowl:IsPredecessorTo?y. ?y ifcowl:RelatingProcess_of_IfcRelSequence?z. }
SELECT ?x ?y {?x ifcwod:isPredecessorTo_IfcProcess ?y}
As proven by benchmarks
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Simplifying query writing
◼ Q1 and Q1’ retrieve all external walls of a building project
◼ Q2 and Q2’ retrieve all doors and their references
April 12th 2016 ifcWOD 14
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Reducing query execution time
Test environment
• Server: Stardog Intel Xeon CPU E5-2430 at 2.2GHz with 2 cores out of 6, 8GB of DDR3 RAM memory
• Client machine Intel Core CPU I7-4790 at 3.6GHz with 4 cores, 8GB of DDR3 RAM memory at 1600MHz
• Building Project: ACTIVe3D (11MB STEP file)
• Queries – each executed 30 times
• Q1, Q2 and Q3 are solely composed of IfcOWL terms
• Q1’, Q2’ and Q3’ are composed of IfcWoD terms
April 12th 2016 ifcWOD 15
Q1 Q1’ Q2 Q2’ Q3 Q3’
Mean (seconds) 0.242 0.026 0.516 0.025 1.348 0.056
Standard Deviation 0.024 0.009 0.019 0.008 0.024 0.017
#Results 37 37 141 141 67 67
Mean Time Reduction (%) 89.26% 95.15% 95.85%
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Reducing data redundancy
April 12th 2016 ifcWOD 16
Modelling data
without IfcWoD
Modelling data
with IfcWoD
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Conclusion and Future Works
Semantic modelling of IFC relations in OWL
Novel modelling that allows an easen application of the Linked (Open) Data principles
Semi-automatic method for ontology conception
IfcWoDlinked to IfcOWL
Simplifies query writing
Improves query response time for retrieving building data
Future works
Analyze the trade-off between data redundancy and query performance
Study the gathering and the inheritance of common properties in the PSDs
April 12th 2016 ifcWOD 17
An
a R
OX
IN –
an
a-m
ari
a.r
ox
in@
u-b
ou
rgo
gn
e.f
rT
rac
isio
ME
ND
ES
DE
FA
RIA
S -
tarc
isio
.me
nd
es
-de
-fa
ria
s@
u-b
ou
rgo
gn
e.f
r
Thank you for your attention
Tarcisio Mendes de Farias – tarcisio.mendesdefarias@checksem.fr
Ana Roxin – ana-maria.roxin@u-bourgogne.fr