Jakob Beetz, Bauke de Vries, Jos van Leeuwen Design Systems group TU/Eindhoven ● Distributed...

42
Jakob Beetz, Bauke de Vries, Jos van Leeuwen Design Systems group TU/Eindhoven Distributed Collaboration in the Context of the Semantic Web Presentation for the 8 th DDSS conference, Heeze

Transcript of Jakob Beetz, Bauke de Vries, Jos van Leeuwen Design Systems group TU/Eindhoven ● Distributed...

Jakob Beetz, Bauke de Vries, Jos van Leeuwen

Design Systems group TU/Eindhoven

● Distributed Collaboration in the Context of the Semantic Web

Presentation for the 8th DDSS conference, Heeze

Overview

● Motivation● Overall system architecture

● DL-based BIM● Query and partial model extraction● Reasoning / inference

● Discussion & Outlook

Traditional Working Methods

● Traditional CA(A)D data is– Non-deterministic

and ambiguous– Episodic– Highly dynamic– Continuous– Does not contain

machine readable knowledge

*

*

**

*

*

*

*

*

*

Central Building Model

● Central Building Model– Founded on

central databases

– No specification for interaction

– Assumes completeness

– Appropriate for smaller projects?

MAS in heterogeneous environments

Local machine / Intranet /Internet

Agent MarketplaceActor Agent

Actor Agent

Simulation wrapper agent

Resource Agent

Knowledge Base

Human DomainExpert

Product Data Base

123

Designer

MAS in heterogeneous environments

I would like to change the size of this roomWill your HVAC unit still fit in?

MAS in heterogeneous environments

I would like to change the size of this roomWill your HVAC unit still fit in?

PDB

Yes but +10dB

Same Specs but max size 2x3x4m ?

MAS in heterogeneous environments

I would like to change the size of this roomWill your HVAC unit still fit in?

PDB

Yes but +10dB

Same Specs but max size 2x3x4m ?

Yes but it’s +10 dB

MAS in heterogeneous environments

I would like to change the size of this roomWill your HVAC unit still fit in?

PDB

Yes but +10dB

Same Specs but max size 2x3x4m ?

Yes but it’s +10 dBRegulations

DB

No

Sound insulationsatisfactory?

MAS in heterogeneous environments

I would like to change the size of this roomWill your HVAC unit still fit in?

PDB

Yes but +10dB

Same Specs but max size 2x3x4m ?

Yes but it’s +10 dBRegulations

DB

No

Sound insulationsatisfactory? Ok, we leave it unchanged

MAS in heterogeneous environments

I would like to change the size of this roomWill your HVAC unit still fit in?

PDB

Yes but +10dB

Same Specs but max size 2x3x4m ?

Yes but it’s +10 dBRegulations

DB

No

Sound insulationsatisfactory? Ok, we leave it unchanged

Yellow pages

UserAgentUserAgent

A

B

C

D

PDB SemWeb Service

SemWeb Service

PDB SemWeb Service

SemWeb Service

PDB SemWeb Service

SemWeb Service

PDB SemWeb Service

SemWeb Service Mapping and reasoning service

Motivation

● Develop semantically enhanced web services, assist Grid computing

● Profit from large set of research, methods and tools developed in other domains

● Use knowledge representation tools for modeling

● Facilitate distribution of information across networks

Performance estimation use case in a MAS

Designer

Domain Expert

Agent Platform

choose aspect tocheck for

«uses»

User agent

Map parameters

«uses»

Query parameters

Yellow Pages

Ontology Service

Search agent

Find cataloguematching country / region

«uses»

«uses»

Get InformationRequired

Infer missingInformation

Prompt for missingInformation

-ask

*

-tell

*

«uses»

«uses»

«uses»

Inference agent

Application wrapper

CalculatePerformance

Handle Query Publish service

«uses»

Advise cases

«uses»

«uses»«uses»

Topological inference task example

IfcWallStandardCase

#949= IFCSPACE(‘guid1',#13,'1',$,$,#947,#939,'big_room',.ELEMENT.,.INTERNAL.,$);

IfcSpace

#97= IFCWALLSTANDARDCASE(‘fooGuid',#13,'Wand-035',$,$,#95,#153,$);

IfcRelSpaceBoundary (as IfcConnectionSurfaceGeometry)???

Chain of agents for the sceneario

CAAD packageUser Agent

Partial Model Extraction

IFC Ontology ifcOWL

ifcOWL Instance Model

Designer

Performance analysis

from simulation

Simulation agent

IFC to ESP-r model converter

ESP-r simulation

ESP-rmodel

Negotiation agent

Reasoner

ifcOWL partial Model with sufficient spatial relations

ifcOWL partial Model with incomplete

spatial relations

Inference rules

ifcOWL partial Model with

incomplete spatial relations

Storing a model as ifcOWL

CAAD packageUser Agent

Partial Model Extraction

IFC Ontology ifcOWL

ifcOWL Instance Model

Designer

Performance analysis

from simulation

Simulation agent

IFC to ESP-r model converter

ESP-r simulation

ESP-rmodel

Negotiation agent

Reasoner

ifcOWL partial Model with sufficient spatial relations

ifcOWL partial Model with incomplete

spatial relations

Inference rules

ifcOWL partial Model with

incomplete spatial relations

Different Knowledge Domains in AEC/FM project

● ERM● Semantic

Networks● Frame

based systems

● DL based systems

Architecture diagram of IFC 2x2

IfcOWL as KRS

• KRS = Terminology (TBox )+ Assertational Knowledge (ABox)

• = (,)

Knowledge Representation System

ReasoningDescription Language

TBox

ABox

Appplication Rules

Basic constructs

Attribute Language

Atomic concepts C,DAtomic Roles R,S

Terminological Axioms:

Equality C D(used to describe definitions)

Inclusion / Supsumption C D(used to describe IS-A realtions)

Basic constructs

Terminological Axioms (cntd.):

Intersection C D

Woman Person Female

Existential quantification R.CMother Woman hasChild.Person

Value Restriction R.CMotherOfDaughtersOnly

Woman hasChild.(Person Female)

Semantic Web Stack

Semantic Web Structure according to Tim Berners-Lee

RDF + RDF-Schema

XML + XML-NS + XML Schema

Unicode

Proof

Logic

OWL Ontology Vocabulary

Trust

URI

Dig

ital

Sig

nat

ure

RDF-based representations

Ora Lassila is the creator of the resource http://www.w3.org/Home/Lassila.

N3 noation: “Ora Lassila” isCreatorOf w3.org/Home/Lassila

<?xml version="1.0"?> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:s="http://description.org/schema/"> <rdf:Description about="http://www.w3.org/Home/Lassila">

<s:Creator>Ora Lassila

</s:Creator> </rdf:Description>

</rdf:RDF>

PredicateSubject Object

Subject ObjectPredicate

IfcOWL T-Box in Protege

IfcOWL A-Box in Protege

IfcOWL

Please find the

Technical specification details of ifcOWL in Gehre, A. Katranuschkov, P. Wix, J. and Beetz, J. (2006).InteliGrid Deliverable D31: Ontology Specification, The InteliGrid Consortium, c/o University of Ljubljana, www.InteliGrid.com.

Extracting partial models

CAAD packageUser Agent

Partial Model Extraction

IFC Ontology ifcOWL

ifcOWL Instance Model

Designer

Performance analysis

from simulation

Simulation agent

IFC to ESP-r model converter

ESP-r simulation

ESP-rmodel

Negotiation agent

Reasoner

ifcOWL partial Model with sufficient spatial relations

ifcOWL partial Model with incomplete

spatial relations

Inference rules

ifcOWL partial Model with

incomplete spatial relations

Partial Model extraction e.g. using SPARQL

Communication

CAAD packageUser Agent

Partial Model Extraction

IFC Ontology ifcOWL

ifcOWL Instance Model

Designer

Performance analysis

from simulation

Simulation agent

IFC to ESP-r model converter

ESP-r simulation

ESP-rmodel

Negotiation agent

Reasoner

ifcOWL partial Model with sufficient spatial relations

ifcOWL partial Model with incomplete

spatial relations

Inference rules

ifcOWL partial Model with

incomplete spatial relations

Classic Agent Model

Agent

Environment

Effector Perceptor

Reasoning

Inter-Agent Communication Layers

Message Transport Protocol

Transport and Signaling

Network Infrastructure

Content Language

Agent Communication Language

Message Envelope

Conversation

TCPUDP

HTTPRMI

FIPA Message Envelop

e

FIPA-ACL

KQML

SLKIF

RDF0

CNPAuction

Communication Layers according to Laamanen & Helin

Topological inference

CAAD packageUser Agent

Partial Model Extraction

IFC Ontology ifcOWL

ifcOWL Instance Model

Designer

Performance analysis

from simulation

Simulation agent

IFC to ESP-r model converter

ESP-r simulation

ESP-rmodel

Negotiation agent

Reasoner

ifcOWL partial Model with sufficient spatial relations

ifcOWL partial Model with incomplete

spatial relations

Inference rules

ifcOWL partial Model with

incomplete spatial relations

Simple Example of DL notation of ER model

– representation of an IFC DoorDoor BuildingElement

OverallHeight.PositiveLength OverallWidth.PositiveLength

SWRL implementation of Rules

fatherOf(?x,?y) ^ brotherOf(?x,?z) -> uncleOf(?z,?y)

SWRL implementation of Rules

ENTITY door;SUBTYPE OF (buildingPart)

height: REAL;WHERE

WR : height > 0;END_ENTITY;

door(?x) ^ hasHeight(?x,?height) ^ swrlb:greaterThan(?height, 0) -> WR1(?x,true)

Thank you

Jakob Beetz

Design Systems group TU/Eindhoven

● Questions, suggestions, comments very welcome

IFC ontology in Protégé

IFC concept structure

Main differences to EXPRESS

● Built-in reasoning capabilities● Built-in functions for creation and

reference of distributed data● Open world vs. closed world

IFC concept structure

● ISO 10303-p28 xml bindings● EXPRESS (ISO 10303-11)

Available Taxonomies and Ontologies

● IFC 2.2x (XML Schema)● Stabu Lexicon (XML Schema)● eCOGNOS (DAML-OIL)● WordNet (OWL)

Semantic Web Layers

RDF Schema example from Richard Vdovjak