SEMANTIC MANAGEMENT OF INTELLIGENT MULTI-AGENTS

SYSTEMS IN A 3D ENVIRONMENT

Florian B

1,2

, Christophe Nicolle

, St

ephane Galland

and Abder Koukam

Laboratoire Electronique, Informatique et Image - UMR CNRS 5258, IUT Dijon Auxerre, Universit

e de Bourgogne

BP 17867 21078 Dijon Cedex, France

Laboratoire Syst

emes et Transports, Universit

e de Technologie de Belfort-Montb

eliard, 90010 Belfort Cedex, France

Keywords:

Ontology, Intelligent multi-agent systems, Knowledge acquisition, Industry foundation classes.

Abstract:

This paper presents a new approach combining the 3D elements composing the environment of mobile agents

with semantic descriptors from Building Information Models. Our proposal is based on the IFC standard,

which is used in the ﬁeld of Civil Engineering to build digital models of buildings during the design phase.

The semantic of IFC objects composing the 3D environment is used to select and set up 3D objects and

elements of simulation scenarios. The result of this process dynamically generates the input ﬁles for the JaSIM

environment that performs the simulation. These ﬁles deserve the representation of the virtual environment in

which the simulation is running. It is represented by two separate ﬁles: a COLLADA ﬁle for the geometry

and a RDF ﬁle for its semantics. Both ﬁles are generated according to the data extracted and selected from an

IFC ﬁle by the user.

1 INTRODUCTION

The construction of a building is organized into sev-

eral steps, from conception to completion. This col-

laborative work requires the involvement of multiple

stakeholders throughout the life cycle of the building.

Many standards have been deﬁned in each trade in-

volved in this life cycle. However, this cooperation

still faces problems of heterogeneity (Vanlande et al.,

2008). To resolve the ﬁrst levels of heterogeneity

(syntactic, structural and schematic) a standard called

Industry Foundation Classes (IFC) was imposed in

the world of civil engineering. The IFC standard was

created in the late 90s by the International Alliance

for Interoperability (now called buildingSMART). Its

goal is to build a common data model for all the ac-

tors in the building industry to resolve problems of

heterogeneity. The IFC standard is the kernel of the

new generation of BIM. Several versions of the IFC

have been published, because of the gradual increase

of the covered area. The heart of the model was stabi-

lized in 2005 and received the ISO certiﬁcation as ISO

/ PAS 16739: 2005. The current release is 2x3TC1.

It is an ASCII ﬁle containing all the elements of the

described building and may be displayed in 3D. The

next version of the IFC, named IFC4, should see its ﬁ-

nal version published at the same time as the interna-

tional standard ISO/IS16739

. IFCs enable the ex-

change of data, either in the form of geometries, but

as objects and their structures (walls, doors, windows,

stairs ...). IFC ﬁles contain a description of all objects

in the buildings and their links. The format also de-

scribes more abstract concepts such as schedules, ac-

tivities, places, organizations, construction cost, etc.

Gradually, publishers of CAD software and gener-

ally all software for civil engineering (structural anal-

ysis, air conditioning ...) develop translation func-

tions of their proprietary language to the IFC standard

(Cruz and Nicolle, 2008). Beyond a simple format for

interoperability, the IFCs, which describe both the ge-

ometric representation of objects in the buildings and

their semantics can be used to manage ontology ag-

gregating all knowledge of the buildings (contractual

document, pictures, dashboards...) (Cruz and Nicolle,

2005; Vanlande et al., 2003; Cruz and Nicolle, 2010).

This semantic translation of building information can

also be used to achieve new goals in qualifying the use

of the building from the design phase. The combined

use of 3D and semantics of IFC is perfectly suited to

the construction of an intelligent multi-agent system

to simulate the behavior of mobile entities in a 3D en-

vironment. This system helps to qualify the use of the

The name of the ISO standard for IFC4

Béhé F., Nicolle C., Galland S. and Koukam A..

SEMANTIC MANAGEMENT OF INTELLIGENT MULTI-AGENTS SYSTEMS IN A 3D ENVIRONMENT.

DOI: 10.5220/0003664300910098

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (KEOD-2011), pages 91-98

ISBN: 978-989-8425-80-5

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

environment and especially to improve, using seman-

tics, the existing processes in the ﬁeld of multi-agent

located in a 3D environment. This paper discusses

an ongoing research on the design of a multi-agent

system based on a semantic indexing of IFC objects.

This indexing process allows to build dynamically an

informed environment (composed of 3D objects se-

mantically indexed) and intelligent agents that can re-

act to the environment according to a context of use.

The new section presents the existing works on us-

ing semantics in Multi-Agents Systems. Section 3

overviews the proposed the principle of the envi-

ronment representation and the architecture of the

semantic-based environment generation tool. Finally,

Section 4 concludes this work and draws several per-

spectives.

2 BACKGROUND

This paper is located in the domain of the simulation

of buildings ﬂows with situated multi-agent systems.

A multi-agent system (MAS) is a system composed of

multiple interacting intelligent software agents (Fer-

ber, 1995). Multi-agent systems can be used to solve

problems that are difﬁcult or impossible for an in-

dividual agent or a monolithic system to solve. A

multi-agent system is situated when the agents are

immersed inside an environment. In the domain of

buildings simulation, an agent is assumed to be a

pedestrian, or any object that owns an autonomous

decision-making process. The environment is then

everything that is not an agent in the buildings.

Three different points of view may be adopted

to study the notion of environment in situated MAS

(Weyns et al., 2007): (i) the part of the system

which is outside the community of the agents; (ii) the

medium for coordination among these agents; or

(iii) the running infrastructure or platform. Weyns

et al. distinguish between the physical environment

and the communication environment (Weyns et al.,

2006). The physical environment provides the laws,

rules, constraints and policies that govern and support

the physical existence of agents and the other entities.

In the rest of this paper, only this aspect of the envi-

ronment is taken.

Several problems may be solved to properly im-

plement an environment: its topological and geomet-

rical description, its dynamics, and the meanings of

each object and zones in the environment. The two

ﬁrst points are addressed by the JaSIM environment

model (Galland et al., 2009). The last point is ad-

dressed by both the integration of semantics in the en-

vironment and modiﬁcation of the agent’s algorithms

that uses them. Most of the approaches found in the

literature are based on the tagging process of the en-

vironment.

• Tagging is Often used as a Kind of Semantic.

The concept of tagging consists in placing some

tags in the environment to inform agents on var-

ious subjects. Basically, tags are considered as

objects placed in the scene, but they do not have

a physical presence. They are invisible for the

viewer of the scene and are only being seen by the

agents. Our proposal includes to describe through

tags the usage of some places, i.e. where an agent

can sit, pass through, etc. Lugrin and Cavazza

(Lugrin and Cavazza, 2007) places various tags

on a single object. These tags are linked between

and can also represent information. For example,

a glass will be represented by a geometry and two

tags. The geometry will deserve its representa-

tion and also dimensions in the simulated world.

The ﬁrst tag is a “containing” tag that will notify

agents that the object on which this tag is applied

can contain some things. The second tag is an

“opening” tag that will represent the fact that the

object is opened, and then the inside of this object

is accessible. These two tags are linked together

to represent the fact that if an agent interact with

the “opening” tag, that will affect the state of the

“containing” one. Finally, a last link is made be-

tween the “opening” tag and the environment to

represent the fact that the opening is accessible di-

rectly from the environment space.

In this proposal, the evolution of an object is done

by modifying the tags and links of this object. For

example, if the glass is clogged, the link between

the opening tag and the environment is deleted

and this tag is no more accessible from the en-

vironment.

Yersin et al. (Yersin et al., 2005) propose to imple-

ment a navigation graph with the help of tagging.

This approach consists in covering the maximum

amount of navigable areas with a minimum num-

ber of discs covering these areas and not overlap-

ping any obstacle. These discs overlap themselves

and form a navigation — center to center — graph

in which agents are sure to do not collide with

any obstacle. Moreover, these discs have labels

to deﬁne the name of the zone in which they are

located. These labels are very useful to select a

target without knowing its position in the environ-

ment.

• Using Roles in Addition of Tagging. In a sim-

ilar way, De Paiva et al. (De Paiva et al., 2005)

propose to put labels on areas in order to asso-

ciate a name with a position. But in opposite to

KEOD 2011 - International Conference on Knowledge Engineering and Ontology Development

the previous works, the agents can only go to a lo-

cation according to its name and not its position.

In addition, the authors propose to assign roles to

the agents and make them evolving in the envi-

ronment according to their roles and the current

simulation time. For example, a kid is at “school”

at 11am and a working adult is at “ofﬁce” at the

same time. This approach is useful to only have

agents that are in a given place, i.e. an agent which

is supposed to be a student is not in the headquar-

ters of a company, except if his behavior needs it.

• Misc. Gutierrez et al. (Gutierrez et al., 2005) pro-

pose to use semantic to describe the interactions

among the agents and the environment objects to

use them with various physical devices (mouse,

keyboard, etc.). This approach can be extended to

describe, for example, the interactions of agents

representing disabled people.

In this paper, the deﬁnition of the environment

is extended with semantics and agent behaviors are

adapted to use these semantic informations. Indeed

with semantics, the result of the simulation of the in-

dividuals is not only based on the geometric features

but also on the information embedded in the spatial

and temporal contexts of the simulation.

3 PROPOSAL

This section presents the principle and architecture of

our proposal, which generates the four JaSIM input

ﬁles from an BIM/IFC ﬁle.

3.1 Previous JaSIM State

Simulation of autonomous entities in a complex urban

system requires dedicated software models. JaSIM

platform (Galland et al., 2009) integrates components

which are required to simulation complex environ-

ments in 1D, 2D and 3D (in particular, particular it is

used for simulation in virtual reality). This platform

integrates several models to reproduce human visual

perception in a virtual environment and endogenous

behavior of this environment. Thus simulated entities

can use the JaSIM platform to perceive and act in a

situated system.

To run a simulation, JaSIM requires two kinds of

input ﬁles. A third kind of ﬁle may be needed if the

simulation has to be displayed.

• SFG File: A SFG ﬁle is a XML-based ﬁle that

can be seen as the scenario description of the sim-

ulation. It will contain various informations about

the simulation and the environment: deﬁnitions of

the places, spawning areas for the agents, goals,

way points, stochastic generation laws, etc. An

example of SFG conﬁguration ﬁle can be seen in

Listing 1.

<? xml v e r s i o n = ” 1.0 ” e n c oding=”UTF−8” ?>

<!DOCTYPE s i m u l a t i o n PUBLIC ” −// s e t . utbm . f r / / DTD

J a S i m C o n f i g u r a t i o n F i l e 3 d v7 . 0 / / EN” ” / f r / utbm / s e t

/ j a s i m / c o n t r o l l e r / c o n f i g / j a s i m −c o n f i g −3d − 7 . 0 . d t d

”>

<s i m u l a t i o n i d = ” 68552 a ab−e71a −44d4−b321 −9d 9 e c 9b937 f a ”

name=”DEMO−SIMULATION”

d a t e = ” 2009−12−23”

a u t h o r s = ”GALLAND S t e phane ”

v e r s i o n =” 0 . 1 ”

d t d v e r s i o n = ” 7 . 0 ”

d e s c r i p t i o n =” S i m u l a t i o n o f P e d e s t r i a n s ”>

<t i m e t y p e =” s t e p ” u n i t = ” m i l l i s e c o n d ” t i m e S t e p = ” 500

” />

<p l a c e i d =” e 19f d4d 1 −73f0 −4283−82b0−0d950d53bb 62

” name= ” Mai n Hal l ”>

<g roun d E n viron m e n t i d =” f 39 772 58 −9d02 −4b71

−8201− b96 d7e 684 e62 ” t y p e =” c o n s t a n t ”>

<i ndoo r G roun d minx=” −128.66 ” miny= ” −118.23 ”

maxx= ” 1 2 8 .66 ” maxy=” 1 1 8 . 2 3 ” z=” 0 ”

s e m a n t i c = ” ” />

</ g r o u ndEnv i r o nmen t>

</ e n v i r o n m e n t>

<spa w ner t y p e = ” a r e a ”

i d = ” 7 a 6d7 aca −4621−4 e f a −97b3−a 5 c 0 8 4 a f f 2 0 f ”

name=”SPAWNER 1 ”

x= ” −8.5209 ”

y= ” −91.81 71 ”

z=” 0 ”

w idth=” 10 ” h e i g h t =” 10” s t a r t A n g l e = ” 0 ” end Ang le

=” 6 . 283 1 8 530 8 ”

p l a c e = ” e 19f d4d 1 −73f 0 −4283−82b0−0d950d 53 bb 62 ”>

<e n t i t y b u d g e t =” 20 ” agen t T ype=” f r . utbm . s e t .

j a s i m . demos . p e d e s t r i a n s . h o lon .

P e d e s t r i a n H o l o n ” >

<f r u s t u m t y p e = ” sp h e r e ” e y e P o s i t i o n =” 1 . 8 ”

f a r D i s t a n c e = ” 10 ” />

</ f r u s t u m s>

<g e n e r a t i o n L a w c l a s s =” f r . utbm . s e t . j a s i m . spawn

. C ons tan tSp awn i ng L aw ”>

</ g e n e r a t i o n L aw>

</ e n t i t y>

</ s p awn er>

</ s p a w n ers>

</ s i m u l a t i o n>

Listing 1: Example of a SFG ﬁle.

• Precomputed Structures of the Environment:

The position of all the static/immobile objects in

the environment are precomputed and saved in-

side a ﬁle containing the corresponding serialized

Java tree. This tree data structure permits to efﬁ-

ciently localize the objects. In the same way a sec-

ond ﬁle may be provided for all the agents which

may be spawned at the start of the simulation. In

both trees, simple semantics can be associated to

objects such as “door”, “window”, etc.

• 3D Model File: The last ﬁle supported by the

SEMANTIC MANAGEMENT OF INTELLIGENT MULTI-AGENTS SYSTEMS IN A 3D ENVIRONMENT

JaSIM platform is optional and is only used when

the simulation is rendered in 3D. It contains all

the geometries of the visible objects which may

be rendered to the ﬁnal used. The format of this

ﬁle should correspond to a standard 3D ﬁle for-

mat such as COLLADA

 or 3D Studio

. All the

geometry provided by this ﬁle may corresponds

to an perceivable object for the agents and previ-

ously described the Java serialized ﬁles.

3.2 JaSIM Evolution

In previous version, JaSIM allowed to represent in-

formation in various ﬁles depending on the kind of

data that needs to be stored. One of the problems that

can be identiﬁed in this kind of representation is that

some piece of information related to the environment

are stored into the scenario ﬁle. For example, places,

for the “place-portal” principle, are stored in the sce-

nario ﬁle despite it is more related to the environment

than the scenario.

One of the goals of the presented work is to split

correctly scenario and environment information. The

SFG ﬁle is thus preserved, but it only contains infor-

mation about simulation’s execution parameters, such

as agents’ types, spawning areas, etc. In addition to

this ﬁle, a Resource Description File (RDF) ﬁle is in-

troduced in order to manage environment’s semantics.

Its geometry is, for its part, supported by the COL-

LADA ﬁle. A reference to IFC elements is kept in the

COLLADA structure in order to retrieve semantics of

the objects in the RDF ﬁle.

3.3 Principle

IFC ﬁles contain a huge amount of information, more

than needed for a multi-agent simulation. The IFC

ﬁles contain a complete description of the building as

illustrated by Figure 1.

Figure 1: Screenshot of an IFC ﬁle. This ASCII ﬁle repre-

sents a complete description of a building in 100838 lines.

That is why it is required to perform a selection on

what piece of information will be extracted from IFC

ﬁles.

The work presented in this paper allows to gen-

erate automatically all ﬁles required by JaSIM to per-

form a simulation. It relies on a kernel-centered archi-

tecture (Section 3.4) which ﬁrst implementation is de-

scribed here. The developed software allows to gen-

erate the four ﬁles that are needed by the JaSIM’s evo-

lution.

The SFG ﬁle only contains informations about the

simulation’s scenario (agents’ spawners, goals, etc.)

as mentioned in Section 3.2. The COLLADA ﬁle

will contain geometry information and, for each ob-

ject, the Global Unique IDentiﬁer (GUID) of the IFC

objects that was used to its generation. The pre-

processed structure that describes mobiles entities is

empty since the presented approach only focuses on

the environment part of MAS. Finally, a RDF ﬁle is

generated containing semantics about extracted IFC

objects. This semantics contains data directly ex-

tracted from the IFC ﬁle and also semantics or con-

straints added by the user. It is linked with the geo-

metrical representation by using the GUID extracted

from the IFC (the same as the one postponed in the

COLLADA ﬁle).

As already mentioned in the beginning of this Sec-

tion, IFC ﬁles contain a lot of information, and MAS

do not need all of them. Our kernel can automatically

ﬁlter these pieces of information relying on an ontol-

ogy schema that will describe what is needed to be

kept for a MAS and transform IFC objects into RDF

elements. Moreover, our kernel also enables the user

to export or not elements in the environment accord-

ing to a speciﬁc context of use. Elements that are not

selected to be exported will thus not be present nei-

ther in the COLLADA ﬁle nor in the RDF ﬁle.

Once this operation is done, the system extracts

geometrical information from the IFC in order to dis-

play to the user the building in a semantic way (i.e.

under a tree form that will correspond to the build-

ing structure: building, storeys, places, etc. as shown

in Figure 2). The 3D representation of the geometry

is also shown to the user. The GUID, stored both in

RDF elements and geometry extracted from IFC ﬁle,

allows to link the tree-representation and the 3D rep-

resentation of the building.

The user can manipulate the RDF elements and to

associate them to SFG concepts or to add constraints

on certain elements or concepts. SFG concepts are,

for most, positioned in a certain location in the envi-

ronment. In this case, the kernel will only use geo-

metrical information extracted from the IFC in order

to place correctly the scenario’s element in the envi-

ronment. In some case, semantics is also used, but

only to set SFG elements’ parameters. For example,

KEOD 2011 - International Conference on Knowledge Engineering and Ontology Development

Figure 2: Example of a semantic representation of an IFC.

there are two kinds of agents’ spawners: the ﬁrst one

spawns agents always on the same point in space and

the second one spawns agent randomly positioned on

a given surface. In this case, the semantics extracted

from the IFC will be used to determine what kind an

agents’ spawner that needs to be used. For example, if

the concept of agents’ spawner is associated to an Ifc-

Space, the spawner will be set as an area spawner that

will be able to spawn agents anywhere in the given

space. In the same manner, if the concept is associ-

ated to an IfcDoor (for example the door that repre-

sents the main entrance) the spawner will only spawn

agents on the exact position of this door.

In addition to performing associations to SFG

concepts, the user can also add some semantic con-

straints to RDF elements. To illustrate this principle,

we use the case of a simulation in which two kinds of

agents can progress, respectively named type A and

type B. The user can, for example, add constraints on

some door to specify that this instance of a door can

only be opened by type A agents. When the simu-

lation will be executed, type A agents will thus see

these doors as crossable elements while type B agents

will see these doors as obstacles. Another example

of constraint is that only type A agents will be al-

lowed to lock or unlock certain instances of doors,

type B agents’ permissions to use the passages will

thus evolve during the run of the simulation. In all

these cases, the COLLADA ﬁle will contain the geo-

metrical representation of the door, and the RDF ﬁle

will contain the fact that it is a door (and all informa-

tion that stem from it such as “ability to open it”, etc.)

and also constraints on these properties, such as the

fact that the door can effectively by opened, but only

by type A agents.

Finally, each action made by the user is stored by a

proﬁle management module. This module will retain,

for each action, the semantic context and the action

of the user. These proﬁles are used to propose to the

user to try to perform automatically actions that the

user could do to set a simulation.

3.4 Architecture

The main part of the proposed software is about the

loading the IFC ﬁle (Figure 3). Two libraries are used.

The ﬁrst one is dedicated to load the IFC structure in

memory according to their speciﬁcations. The sec-

ond one loads the geometry in a directly usable shape

structure and not in geometry description that need to

be processed.

A central kernel was developed to manage and

launch these two libraries, and link them to get the

shape representation of a given IFC element from the

speciﬁcation-compliant memory representation. This

kernel also manages the generation of the JaSIM sce-

nario elements. It also allows communication with the

Graphical User Interface (GUI) to update links and

other data according to user actions.

This kernel is not dependent on the GUI and can

thus be used with another GUI or other user interac-

tion manner. The application structure is shown on

Figure 4. As shown on mentioned Figure, the kernel

loads the IFC ﬁle in order to extract IFC objects and

geometry that are put respectively in the display tree

and three-dimensional display. The developed GUI

allows to interact with the IFC structure or selection

and only calls kernel functionalities. It can thus be

easily replaced by other user interaction methods.

To perform the JaSIM ﬁle generation, the kernel

uses ﬁle generating modules shown in Figure 5. The

COLLADA

 export module has been developed in

the scope of being reusable in any application. It

takes as input a Java3D scene graph (built from the

IFC ﬁles) and generate the COLLADA

 schema that

corresponds to the given structure. Each Java3D el-

ement can be associated to some extra data that al-

lows us to keep a relation between the IFC objects

and the purely geometrical objects, i.e. GUID of IFC

objects is postponed in the COLLADA structure. The

SFG (JaSIM Scenario Conﬁguration) generator mod-

ule parses the IFC data structure of our kernel to re-

trieve associations to SFG concepts and uses IFC ele-

ments’ description to get the geometrical position and

then generate the SFG ﬁle.

Finally, proﬁle management is done by the ker-

nel. It updates the proﬁle at each selection and re-use

the proﬁle on the IFC loading to perform associations.

The proﬁle update process is illustrated on Figure 6.

SEMANTIC MANAGEMENT OF INTELLIGENT MULTI-AGENTS SYSTEMS IN A 3D ENVIRONMENT

Figure 3: Global chart of our proposal.

Figure 4: Kernel centered architecture of our proposal.

Figure 5: File generation process.

4 CONCLUSIONS & FUTURE

WORKS

This paper discusses an ongoing research on the de-

sign of a multi-agent system based on a semantic in-

dexing of IFC objects. This paper presents the use of

IFC ﬁles to generate a MAS environment. These ﬁles

Figure 6: Proﬁle management process.

are based on objects’ description merging semantics

and 3D geometries. Our work is done in order to test

the viability of the usage of IFC ﬁles in MAS simula-

tion domain.

Using IFC ﬁles as a starting point in MAS simula-

tion is beneﬁcial according to the quality of the data,

which are always up-to-date. A simulation can be ex-

ecuted as soon as the building is designed to test the

quality and the level of compliance of such design.

Moreover, using BIM as a MAS input also enables

to bring a high level of semantic information to the

environment that can be used by the agents in turn.

The focus of this paper is on the semantic analysis

of IFC objects composing the 3D environment. It is

used to select and set up 3D objects and elements of

simulation scenarios. The result of this process gener-

ates semi-automatically and dynamically input ﬁles to

the JaSIM environment that performs the simulation

at the end. Our next goal is to improve our software to

get a better automation in the selection process. We

are developing an ontology-based IFC. The ﬁrst re-

KEOD 2011 - International Conference on Knowledge Engineering and Ontology Development

Figure 7: Example of RDF representation of IFC Object.

sults of this work are the translation of the building

objects into COLLADA and RDF ﬁles as depicted in

Figure 7.

This ontology will be used to describe associa-

tions to support more parameters and build rules, in-

cluding the management of the context. For exam-

ple, in our environment, doors and windows are both

opening elements and these elements can be seen by

the agents as crossable elements to get out from a

room. Nevertheless, is a window still a valid exit if

this window is on the 6th ﬂoor? Rules will make it

possible to perform a better classiﬁcation of the ele-

ments and to apply several restrictions and ﬁlters on

the associations. Finally, this architecture will help

to apply ﬁne associations and to improve the environ-

ment, or at least the simulation scenario.

ACKNOWLEDGEMENTS

This work is funded by the region of Franche Comt

and receive grants from a cooperative project between

the Franche Comt

e and the Bourgogne. Thanks to the

Checksem and SeT teams and Jordan Simonot for his

help in the COLLADA export part of this project.

REFERENCES

Cruz, C. and Nicolle, C. (2005). 3d reconstruction based

on semantic information for architectural applications.

Schriftenreihe Informations- und Messtechnick.

Cruz, C. and Nicolle, C. (2008). Cad software and inter-

operability. Encyclopedia of Information Science and

Technology Second Edition (8-Volume Set).

Cruz, C. and Nicolle, C. (2010). Active3d: Semantic and

multimedia merging for facility management. 6th In-

ternational Conference on Web Information Systems

and Technologies.

De Paiva, D., Vieira, R., and Musse, S. (2005). Ontology-

based crowd simulation for normal life situations. In

Computer Graphics International 2005, pages 221–

226. IEEE.

Ferber, J. (1995). Les Syst

emes Multi-Agents : vers une

intelligence collective. InterEditions.

Galland, S., Gaud, N., Demange, J., and Koukam, A.

(2009). Environment model for multiagent-based

simulation of 3d urban systems. In the 7th Euro-

pean Workshop on Multi-Agent Systems, Ayia Napa,

Cyprus.

Gutierrez, M., Thalmann, D., and Vexo, F. (2005). Seman-

tic virtual environments with adaptive multimodal in-

terfaces. In Multimedia Modelling Conference, 2005.

MMM 2005. Proceedings of the 11th International,

pages 277–283. IEEE.

Lugrin, J. and Cavazza, M. (2007). Making sense of virtual

environments: action representation, grounding and

common sense. In Proceedings of the International

SEMANTIC MANAGEMENT OF INTELLIGENT MULTI-AGENTS SYSTEMS IN A 3D ENVIRONMENT

Conference on Intelligent User Interfaces, Honolulu,

Hawaii, USA, pages 225–234.

Vanlande, R., Cruz, C., and Nicolle, C. (2003). Managing

ifc for civil engineering projects. In Proceedings of the

twelfth international conference on Information and

knowledge management, pages 179–181. ACM.

Vanlande, R., Nicolle, C., and Cruz, C. (2008). IFC and

building lifecycle management. Automation in Con-

struction, 18(1):70–78.

Weyns, D., Ominici, A., and Odell, J. (2007). Environ-

ment as a ﬁrst-class abstraction in multi-agent sys-

tems. Journal on Autonomous Agents and Multi-Agent

Systems, 14(1):5–30.

Weyns, D., Parunak, H., Michel, F., Holvo

et, T., and Ferber,

J. (2006). Environments for multiagent systems state-

of-the-art and research challenges. In Third Interna-

tional Workshop E4MAS, volume 4389, pages 1–47.

Springer.

Yersin, B., Maım, J., de Heras Ciechomski, P., Schertenleib,

S., and Thalmann, D. (2005). Steering a virtual crowd

based on a semantically augmented navigation graph.

In Proc. The First International Workshop on Crowd

Simulation (V-CROWDS’05), Lausanne, Switzerland,

pages 169–178. Citeseer.

KEOD 2011 - International Conference on Knowledge Engineering and Ontology Development