ugent research projects on linked data in architecture and construction

UGent Research Projectson Linked Data

in Architecture and Construction

Presentation Technion Haifa

18 January 2017

Prof. Dr. Ir.-Arch. Pieter Pauwels

Ghent University, Department of Architecture and Urban Planning

UGent SmartLab

Ghent University

Faculty of Engineering and Architecture

Department of Architecture and Urban Planning

UGent SmartLab

Prof. Ronald De Meyer

Prof. Pieter Pauwels

Dr. Ruben Verstraeten

Dr. Tiemen Strobbe

Mathias Bonduel

Willem Bekers

Sebastiaan Leenknegt

Nino Heirbaut

3

Pieter Pauwels

• 2003-2008: Ba-Ma Civil Engineering - Architecture (UGent)

BIM

• 2008-2012: PhD Civil Engineering - Architecture (UGent)

BIM -> SemWeb

• 2012-2014: Postdoc University of Amsterdam (UvA)

• 2014-2017: Postdoc Ghent University

SemWeb + BIM

4

Current developments and commitments

- Linked Data in Architecture and Construction (LDAC) workshops• 2012: Ghent• 2014: Helsinki• 2015: Eindhoven• 2016: Dijon

- W3C Community Group on Linked Building Data (LBD)• BOT ontology• use cases that rely on combination of datasets

- linked data working group (LDWG) within BuildingSMART International• ifcOWL ontology

=> STANDARDISATION + APPROPRIATE USAGE OF STANDARDS

6

Outline

1. What is Linked Data? What are Semantic Web technologies?

2. The standards: buildingSMART and W3C

3. Research projects

7

The cool and awesome intro movies

https://vimeo.com/36752317

https://www.youtube.com/watch?v=4x_xzT5eF5Q

https://www.youtube.com/watch?v=OM6XIICm_qo8

https://vimeo.com/36752317

https://www.youtube.com/watch?v=4x_xzT5eF5Q

https://www.youtube.com/watch?v=OM6XIICm_qo

Linked Open Data cloud (LOD)

http://tomheath.com/blog/2009/03/linked-data-web-of-data-semantic-web-wtf/9

http://tomheath.com/blog/2009/03/linked-data-web-of-data-semantic-web-wtf/

• RDF stands for Resource Description Framework

• RDF is a standard data model for describing web resources– Note: ‘web resources’ can make statements about anything in the real

world: DBPedia, geography, building information, sensors, … anything goes

• RDF is designed to be read and understood by computers

• RDF is not designed for being displayed to people

• RDF is written in XML

• RDF is a W3C Recommendation

http://www.w3schools.com/webservices/ws_rdf_intro.asp

easily used

usually

-> standardisation

not a file format, not a syntax, not a schema, … => a data model

RDF??

10

http://www.w3schools.com/webservices/ws_rdf_intro.asp

LABELLED

DIRECTED

Triple

RDF Graphs, what are they?

11

RDF graphs are DIRECTED, LABELLED GRAPHS

RDF Graphs, what are they not?Hierarchies (cfr. XML)

Relational databases (cfr. SQL)

12

RDF Data Model

predicatesubject object

13

Connecting Triples

SUBJECT OBJECTPREDICATE

OBJECT

PREDICATE

OBJECT

PREDICATE

OBJECTPREDICATE

14

The result: an RDF graph

15

https://www.w3.org/DesignIssues/diagrams/sweb-stack/2006a.png


@prefix b: <http://www.today.net/building#> .

@prefix c: <http://www. today.net/city#> .

<http://www.today.net/today#building_1>

b:hasRoom <http://www. today.net/today#room_1> ;

b:hasName “Virtual Construction Lab";

c:partOfCity <http://cities.com/haifa> .

<http://cities.com/haifa>

c:inCountry <http://cities.com/israel> ;

c:hasName “Haifa” .

Example RDF graph

17

• URI stands for Uniform Resource Identifier

• Purpose: Obtain globally unique identifiers, so that information can be exchanged globally.

• Structure:

<http://www.today.net/today#building_1>Namespace Name

Uniform Resource Identifiers (URIs)

18

URI

URI

URI

URI

URI

URIURI

URI

URI

URI

URI

19

MyBuilding Cities

Data integration over the web is now possible

20

• distributed / decentralisedinformation management

• interactive information search and reasoning over the web

• sharing partial data

Main principles

21

Linked Open Data cloud (LOD)



Ontologies



rvt:hasGirder

rvt:hasSlab

rvt:Corbel

rvt:Girder

rvt:Column

rvt:Slab

rvt:InternalBeam

COL_001

rdf:type

rvt:hasCorbel

rvt:hasGirder

rvt:hasSlab COR_001

GIR_001

COR_002

COL_002

rdf:type

rvt:Column rvt:Column

rdf:type rdf:type

rdf:type

rvt:Girder

rvt:Corbel rvt:Corbel

rvt:Slab

rvt:hasCorbel rvt:hasCorbel

rvt:hasGirder rvt:hasGirder

Basic schema of the ontology: Instance sample:

SLAB_1 SLAB_2 SLAB_3 SLAB_4 SLAB_5

rvt:hasSlab

rdf:type

rvt:hasInternalBeam

G. Costa and P. Pauwels. Building product suggestions for a BIM model based on rule sets and a semantic reasoning engine. Proceedings of the 32nd CIB W78 Conference on Information Technology in Construction 2015. pp 98-107.

BIM

GIS

BEMS

sensor

FM

no full integrationrather on-demand high-quality information exchange

regulations

29

Bring BIM into the Semantic Web

BIM30

http://www.buildingsmart-tech.org/future/linked-data/

31

LDAC 2015

LDAC 2014

LDAC 2012

32

Joining / combining initiatives

W3C LBD Community Group BuildingSMART Linked Data Working Group

linkedbuildingdata.net

www.w3.org/community/lbd/

ifcOWL

linkedbuildingdata people

LDAC event

bSDD

MVD

33

Outline




37

Standardisation bodies

CEN/TC 442

ISO TC59

Linked Data WGOpenBIMGuides WGBuildingSMART Benelux

38

BuildingSMART Standards Summit Jeju, Korea25 - 29 September 2016

ISO TC/59 Plenary WeekBerlin, Germany4 - 11 October 2016

CEN TC 442 WG meetingsBerlin, Germany12 - 13 September 2016

39

buildingSMART standardisation strategy

bSI

S t a n d a r d i s a t i o n

ISO

CEN

National Standardshttp://buildingsmart.org/

http://buildingsmart.org/

The buildingSMART triangle



Fit in BuildingSMART activities

http://www.buildingsmart.org/standards/technical-vision/technical-roadmaps/

http://www.buildingsmart.org/standards/technical-vision/technical-roadmaps/

SingaporeITM

October2015

RotterdamISM April2016

LDAC 2015Eindhoven

CIB W78 2015

Eindhoven

LDAC 2014Helsinki

SWIMingVoCamp

2016Dublin

LDAC 2016Madrid

Toronto ITM

October2014

WatfordITM March

2015

KoreaISM

September2016

SWIMingVoCamp

2016London

44

Image courtesy: Jakob Beetz, TU Eindhoven

IFC

INFRA

SENSOR

GIS

45

Aims:

1. ifcOWL ontology

2. align with buildingSMART efforts

3. LD-oriented support

46

EXPRESSIFC-SPF

XSDXML

ifcOWLRDF

Pieter Pauwels and Walter Terkaj, EXPRESS to OWL for construction industry: towards a recommendable and usable ifcOWL ontology. Automation in Construction 63: 100-133 (2016).

conversion procedure EXPRESS schema to OWLIFC

Schema

Simple data type

Defined data type

Aggregation data typeSET data type --------

LIST & ARRAY data type --------

Constructed data typeSELECT data type --------

ENUMERATION data type --------

Entity data typeAttributes --------

Derive attrWHERE rules

FunctionsRules

ifcOWLOntology

owl:class + owl:DatatypeProperty restriction

owl:class

owl:class-------- non-functional owl:ObjectProperty-------- indirect subclass of express:List

owl:class-------- rdfs:subClassOf for owl:classes-------- rdf:type for owl:NamedIndividuals

owl:class-------- object properties

----

Pieter Pauwels and Walter Terkaj, EXPRESS to OWL for construction industry: towards a recommendable and usable ifcOWL ontology. Automation in Construction 63: 100-133 (2016).

ifcOWL ontologies available

Ifc2x_all_lf.expIFC2X2_ADD1.expIFC2X2_FINAL.exp

IFC2X2_PLATFORM.expIFC2X3_Final.expIFC2X3_TC1.exp

IFC4.expIFC4_ADD1.exp

not supportednot supportednot supportednot supportedIFC2X3_Final.owl / .ttlIFC2X3_TC1.owl / .ttlIFC4.owl / .ttlIFC4_ADD1.owl / .ttl

http://ifcowl.openbimstandards.org/IFC4_ADD1http://ifcowl.openbimstandards.org/IFC4

http://ifcowl.openbimstandards.org/IFC2X3_Finalhttp://ifcowl.openbimstandards.org/IFC2X3_TC1

52

http://ifcowl.openbimstandards.org/IFC4_ADD1


http://w3id.org/ifc/IFC4


http://w3id.org/ifc/IFC2X3_Final


http://w3id.org/ifc/IFC2X3_TC1

53

HTML documentation pages

InfrastructureRoom

Technical Room

Building Room

Product Room

RegulatoryRoom

BuildingSMART

BIMInfra GIS

IDMsMVDsBIM-

Guides

bSDD RulesifcOWL

54

55Jakob Beetz, Henk Schaap, Pieter Pauwels, and Jim Plume. Linked Data for Infrastructure.

Presentation at BuildingSMART Int’l Standards Summit 2016, Jeju, Korea.

56Ana Roxin, Pieter Pauwels. Reasoning with rules - Applications to (1) N3/EYE and (2) Stardog.

Presentation at BuildingSMART Int’l Standards Summit 2016, Jeju, Korea.

Image from: Lars Bjørkhaug. Integration of bSDD into the IfcDoc tool. Presentation at BuildingSMART Int’l Standards Summit 2016, Jeju, Korea.

IFC-SPF

EXPRESS

MVD subset

MVDxml

SimpleQueryAccess

GAP

SimpleBIM BIMSPARQL

Pieter Pauwels, Ana Roxin. SimpleBIM: from full ifcOWL graphs to simplified building graphs. Proceedings of the 11th ECPPM Conference, pp. 11-18, 2016, Limassol, Cyprus.

Chi Zhang and Jakob Beetz. Querying Linked Building Data Using SPARQL with Functional Extensions. Proceedings of the 11th ECPPM Conference, pp. 11-18, 2016, Limassol, Cyprus.

Outline




1. Compliance checking

2. IFC to X3D to STL (and back)

3. Query and reasoning performance benchmark

4. SimpleBIM

5. Linked Data in Infra

60

SOURCE: http://neo4j.com/61

Logics: overview

First Order Logic (FOL)

Second Order Logic (SOL)

Horn Logic

Datalog

Propositional Logic

Non-monotonic Logic (NML) Defeasible Reasoning

Monotonic Logic

Predicate Logic

Description Logic (DL)

subsets

N3

SWRL

Prolog

64

Monotonic vs. Non-monotonic logic

Non-monotonic Logic (NML) Defeasible Reasoning

Monotonic Logic

Retraction of inferences in the light of new information

Inferences are guaranteed, also when new information is added

65

Order, order!


Second Order Logic (FOL)

Propositional Logic

Variables quantify over individuals and relations

Variables quantifyover individuals

No variables or quantifiers

Predicate Logic

66

FOL subsets: tastes of logic


Horn Logic

Datalog

Predicate Logic

Description Logic (DL)

subsets

SWRL

N3

subsets

Prolog

OWL

67

1/5: COMPLIANCE CHECKINGPieter Pauwels, Ghent University

Ana Roxin, Université de Bourgogne

Abox – Tbox – Rbox

ABox

TBox

RBox

Instances

Ontology

IF-THEN rules

69

Korean Building Authority (KBA) regulations

• A stair is connected to an object having an exit to ground floor

• The distance from the stair to the exit is not greater than 30000IF

• The stair is a valid exitTHENPREFIX kba: <http://koreanbuildingcode.org/KR-BA-34-01/>PREFIX math: <http://www.w3.org/2000/10/swap/math#>PREFIX add: <http://www.additionalelements.org/>

IF {?s add:isConnectedToStair ?obj .?obj kba:hasExitOnGroundFloor "true" .?s kba:hasEscapeDistanceToStaircase ?value .?value math:notGreaterThan 30000 .

}

THEN {?s kba:isValid "true" .

}

Ana Roxin, Pieter Pauwels. Reasoning with rules - Applications to (1) N3/EYE and (2) Stardog. Presentation at BuildingSMART Int’l Standards Summit 2016, Jeju, Korea.

Reasoning with the EYE and Stardog reasoner

inference engine

OWL ontologies

query

User

RDF Repository

interface

IF-THEN rule repository

response in RDF graph

EYE reasoningengine

N3 OWLRDF

SPARQL

RDF / CSV

English


RDF graphLog:implies(IF THEN)

N3Logic

@prefix kba: <http://koreanbuildingcode.org/KR-BA-34-01/> .@prefix add: <http://www.additionalelements.org/> .@prefix math: <http://www.w3.org/2000/10/swap/math#> .

{?s add:isConnectedToStair ?obj .?obj kba:hasExitOnGroundFloor "true" .?s kba:hasEscapeDistanceToStaircase ?value .?value math:notGreaterThan 30000 .

}=>{?s kba:isValid "true" .

} .


Serialisations of RDF graphs

https://www.w3.org/DesignIssues/diagrams/n3/venn

https://www.w3.org/DesignIssues/diagrams/n3/venn

Rule-checking scenario

• 2 repositories

• Facts1.ttl + ont.ttl + rs1.ttl

• Facts2.ttl + ont.ttl + rs1.ttl

• SPARQL queries addressing the properties being impacted by the rules in the rule set (rs1.ttl)


Inference: rule 1

Query 1:

Output facts1:

Output facts2:


Inference: rule 2

Query 2:

Output facts1 and facts2:


Inference: rule 3

Query 3:

Output facts1:

Output facts2:


Inference: rule 4

Query 4:

Output facts1:

Output facts2:


Inference: rule 5

Query 5:

Output facts1 and facts2:


Inference: rule 6

Query 6:

Output facts1:

Output facts2:


Inference: rule 7

Query 7:

Output facts1:

Output facts2:


2/5: IFC TO X3D TO STL (andback)Pieter Pauwels, Davy Van Deursen, Jos De Roo, Tim Van Ackere, Ronald De Meyer, Rik Van de Walle and Jan Van Campenhout

Ghent University

3/5: QUERY AND REASONING PERFORMANCE BENCHMARKPieter Pauwels, Tarcisio Mendes de Farias, Chi Zhang, Ana Roxin, Jakob Beetz, Jos De Roo, Christophe Nicolle

99

Performance benchmark variables

Schema (TBox)

• ifcOWL

Instances (ABox)

• 369 ifcOWL-compliantbuilding models

Rules (RBox)

• 68 data transformation rules

101

• Implemented based on the open source APIs of Topbraid SPIN (SPIN API 1.4.0) and Apache Jena (Jena Core 2.11.0, Jena ARQ 2.11.0, Jena TDB 1.0.0)

• Rules are written with TopbraidComposer Free version, and they are exported as RDF Turtle files.

• A small Java program is implemented to read RDF models, schema, rules from the TDB store and query data.

• All the SPARQL queries are configured using the Jena org.apache.jena.sparql.algebrapackage

• To avoid unnecessary reasoning processes, in this test environment only the RDFS vocabulary is supported.

SPIN + Jena TDB

• Version ‘EYE-Winter16.0302.1557’ (‘SWI-Prolog 7.2.3 (amd64): Aug 25 2015, 12:24:59’).

• EYE is a semi-backward reasoner enhanced with Euler path detection.

• As our rule set currently contains only rules using =>, forward reasoning will take place.

• Each command is executed 5 times

• Each command includes the full ontology, the full set of rules and the RDFS vocabulary, as well as one of the 369 building model files and one of the 3 query files.

• No triple store is used: triples are processed directly from the considered files.

EYE

• 4.0.2 Stardog semantic graph database (Java 8, RDF 1.1 graph data model, OWL2 profiles, SPARQL 1.1)

• OWL reasoner + rule engine.

• Support of SWRL rules, backward-chaining reasoning

• Reasoning is performed by applying a query rewriting approach (SWRL rules are taken into account during the query rewriting process).

• Stardog allows attaining a DL-expressivity level of SROIQ(D).

• In this approach, SWRL rules are taken into account during the query rewriting process.

Stardog

102

Queries

• We have built a limited list of 60 queries, each of which triggers at least one of the available rules.

• As we focus here on query execution performance, the considered queries are entirely based on the right-hand sides of the considered rules.

• 3 queries: Query Query Contents

Q1 ?obj sbd:hasProperty ?p

Q2?point sbd:hasCoordinateX ?x .?point sbd:hasCoordinateY ?y .?point sbd:hasCoordinateZ ?z

Q3 ?d rdf:type sbd:ExternalWall

103

Results• Queries applied on 6 hand-picked

building models of varying size

• In the SPIN approach• For Q1 and Q2, the execution time =

backward-chaining inference process + actual query execution time

• For Q3, execution time = query execution time itself

• In the EYE approach• Networking time is ignored

• In the Stardog approach• Execution time = backward-chaining

inference + actual query execution time

QueryBuildingModel

SPIN (s)

EYE (s)

Stardog (s)

Q1(simple,

littleresults)

BM1 135,36 37,11 13,44

BM2 1,47 0,29 0,17

BM3 24,01 4,87 1,4

BM4 41,28 12,95 3,55

BM5 4,99 1,05 0,33

BM6 0,55 0,16 0,08

Q2(simple,

manyresults)

BM1 46,17 2,10 6,82

BM2 92,03 4,20 15,83

BM3 82,68 4,12 15,28

BM4 19,93 1,04 2,81

BM5 3,69 0,21 1,36

BM6 0,74 0,045 1,00

Q3(complex)

BM1 0,001 0,001 0,07

BM2 0,006 0,003 0,12

BM3 0,002 0,003 0,31

BM4 0,005 0,001 0,20

BM5 0,006 0,013 0,20

BM6 0,001 0,001 0,13

104

Query time related to result count• For Q1 for each of the considered

approaches

• (green = SPIN; blue = EYE; black = Stardog)

• For Q2 for each of the considered approaches

• (green = SPIN; blue = EYE; black = Stardog)

105

Findings

Impact on performance from many factors, in order of impact:

1. Indexing algorithms, query rewriting techniques, and rule handling strategies

2. Forward- versus backward-chaining

3. Type of data in the building model

4. Storage in the triple store

5. Number of output results

106

4/5: SIMPLEBIM

107

Pieter Pauwels, Ghent University

Ana Roxin, Université de Bourgogne

Pieter Pauwels, Ana Roxin. SimpleBIM: from full ifcOWL graphs to simplified building graphs. Proceedings of the 11th European Conference on Product and Process Modelling. p.11-18.

T. Liebich. buildingSMART Data Standards. BuildingSMART International Summit 2012.

ISO 29481

ISO 16739

IFC, MVDs and IDM

MVDusability

110

SimpleBIM

IFC-SPF

EXPRESS

MVD subset

MVDxml

SimpleQueryAccess

GAP

SimpleBIM BIMSPARQL

Pieter Pauwels, Ana Roxin. SimpleBIM: from full ifcOWL graphs to simplified building graphs. Proceedings of the 11th European Conference on Product and Process Modelling. p.11-18.

Chi Zhang, Jakob Beetz. Querying Linked Building Data Using SPARQL with Functional Extensions. Proceedings of the 11th European Conference on Product and Process Modelling.

RDFIFC-SPF

ifcOWLEXPRESS

RDF

simpleBIM

Converter?Rules?

…?

Converter?Rules?

…? 113

115

inst:IfcWindow_1893 inst:IfcWindow_1842

inst:IfcWallStandardCase_696

simplebim:hasWindow simplebim:hasWindow

Statistics of the test file

• File size: 767kB

• Triple count: 10,173 distinct

• Class instances: 4222 (5535)• 233 / 4222 ifcowl:IfcRelationships• 686 / 4222 list:OWLList• 417 / 686 ifcowl:IfcLengthMeasure_List• 764 / 4222 expr:STRING

117

Simplification strategy

118

1• Removing geometric information

2• Unwrapping data types

3• Rewriting properties

4• IfcRelationship instances

Simplifying IfcRelationship instances

119

Simplifying IfcRelationship instances

120

Unwrapping data types

121

Removing geometric information

122

Rewriting PSETs and property values

123

124

Rewriting PSETs and property values

Results (1)

125

1. Removal of geometric information

• 10,173 triples to 6,927 triples

• 767kb to 476kb

• 31% (file size) – 38% (triple count)

2. Unwrapping data types

• 3,897 triples

• 279kb


Results (2)

126

3. Rewriting properties

• 1,630 triples

• 112kb


4. IfcRelationship instances

• 1,339 triples

• 83kb


Results (3)

127

Model File size Triple count

ifcOWL simpleBIM ifcOWL simpleBIM

1 767kb 83kb 10 173 1 339

2 16,7MB 1029kb 225 135 16 836

Average reduction of 91,58% Average reduction of 89%

REDUCTION TO:8,5% of file size10,3% of triple count

5/5: LINKED DATA IN INFRA

128128

Jakob Beetz, Henk Schaap, Pieter Pauwels, Jim Plume

T. Liebich (2013), IFC for Infrastructure, INFRA-BIM Workshop, Helsinki

Jakob Beetz, GIS / BIM interoperabiliteit: STUMICO presentatie April 2014. http://www.slideshare.net/JakobBeetz/gis-bim-interoperabiliteit-stumico-presentatie-april-2014

http://www.slideshare.net/JakobBeetz/gis-bim-interoperabiliteit-stumico-presentatie-april-2014

Jakob Beetz, Michelle Lindlar, Stefan Dietze, Ujwal Gadiraju, Dag Field Edvardsen, Lars Bjørkhaug, OntologicalFramework for a Semantic Digital Archive. DuraArk Deliverable D3.3.2.

132

Infra as Linked Data – courtesy of Jakob Beetz

Outline




1. Compliance checking

2. IFC to X3D to STL (and back)

3. Query and reasoning performance benchmark

4. SimpleBIM

5. Linked Data in Infra

134

Thank you

Pieter Pauwels

[email protected]