ONTOLOGICAL CONFIGURATOR

A Novel Approach

F. Clarizia, F. Colace and M. de Santo

DIIIE-Università degli Studi di Salerno, Via Ponte Don Melillo, 1 84084 Fisciano (Sa), Italy

Keywords: Ontology, Product Configurator.

Abstract: The ability to create customized product configurations which satisfies a user’ needs is one of the major

aims desired by companies. In particular, the design and implementation of web-based services that allows

users to create and customize in a simple and intuitive way the desired product. One research area that has

lacked progress is the definition of a common vocabulary that enables consumer-to-manufacturer and

manufacturer-to-manufacturer communication. Enabling this communication opens possibilities such as the

ability to express customer requirements correctly and to exchange knowledge between manufacturers. A

popular approach to express knowledge uses ontological formalisms. Using ontology, a vocabulary can be

defined that allows interested parties to specify and share common knowledge and serves to define a

framework for the representation of the knowledge. With this aim, this paper presents an ontology-based

configurator system that finds the configuration that maximizes the user’s needs starting from the desired

requirements, the available components, the context information and previous similar configurations. The

process by which the system finds the candidate configuration follows an approach known as “Slow

Intelligence”. This paper presents in detail the proposed approach and presents its first application.

1 INTRODUCTION

In the past, companies relied on standard

configurations that reduced production costs while

increasing profits. However, as new manufactures

join the market, existing ones must find ways to

attract new or retain existing customers. One

method that is gaining popularity is the ability to

personalize products to a customer’s need. When

users are able to obtain a product that meet their

needs as opposed to most closely meeting their

needs, their perception of the manufacturer

increases. This increase in perception allows

manufacturers to gain repeat customers resulting in

an increase of their profits. However, the ability to

generate a customized product that meets a user’s

needs is still challenging for most companies.

Customers ask for personalized products prompting

companies to consider mass customization. Mass

customization brings a change in how product

design is organized affecting the cost-efficiency of

mass production. Many researchers believe that

product configuration is an effective answer to this

problem (Ding, 2008). In this scenario, a customer is

able to experiment any kind of product

configuration, which is controlled from the point of

view of its technical feasibility and cost. This

approach is effective not only for products but also

in the world of services: in fact, customer often

requests a complex service starting from a portfolio

of base services (insurance services, travel services

and so on). Therefore, the product configurator is

becoming the killer application for the management

of companies’ business processes: In fact, the

customer’s order drives the full chain of processes

(Felfernig, 2002). The configurator, however, has

also to furnish the best solution for the customer and

check the actual feasibility of the product. In other

words, a good configurator aims to be the perfect

seller being able to satisfy customers’ needs 24

hours per day. Product development based on

customer preferences is key to obtaining larger

market share and faster sales growth for

organizations. The Internet has fostered the creation

of e-businesses and the building of a real, interesting

and distributed market. Many corporations have

introduced on their websites an interactive product

configurator allowing customer to experiment with

options and be supported in the product selection

process (Franke, 2003). Even though this issue

represents a very challenging and interesting

research topic, there is a lack of research activities.

Current configurator implementations include

515

Clarizia F., Colace F. and de Santo M. (2010).

ONTOLOGICAL CONFIGURATOR - A Novel Approach.

In Proceedings of the 12th International Conference on Enterprise Information Systems - Information Systems Analysis and Speciﬁcation, pages

515-520

DOI: 10.5220/0002976505150520

 SciTePress

simple rule-based system, a hierarchical

questionnaire or a passive system that collects with

no logic, the preferences of the customer. In these

cases, the product configurator is just a product

viewer for the customer and does not implement any

reasoning logic or user adaptive approach.

Customers’ requests are passively received without

further reasoning requiring the customer to be

familiar with both the product structure and its

functions and so often the final configuration is not

the best for him. Recently, there have been some

proposals in literature (Ding, 2008) (Park, 2008)

(Youliang, 2007) for the introduction of an

intelligent configurator. They propose an ontology

based approach which seems to be an effective

methodology for the improvement of the actual

configurator (Yang, 2008). In this paper, a

framework for the assisted product configuration,

based on the use of ontology formalism and

methodologies, is proposed. It works mainly by the

use of four different ontologies: the functionality

ontology obtained by the analysis of the user’s

request, the component’s ontology obtained by the

support of an expert, the customer’s context

approach and the ontology of previous

configuration. In particular, the proposed

configurator follows a Slow Intelligent System’s

model. These models are general-purpose systems

characterized by being able to improve their

knowledge over time using the working context,

expert knowledge and task methodologies. A Slow

Intelligent System is one that given a particular task

is able to reason and provide an answer after

completing a process of enumeration, elimination

and concentration and continuously learns, searches

for new knowledge and shares experience with other

peers. In a Slow Intelligent System, the information

is represented by the use of ontology formalism. In

the following paragraphs, this approach will be

detailed. This paper follows this structure: the next

section describes the Slow Intelligent model

approach. The third section explains in details the

proposed configurator and a working example is

furnished. Finally, conclusions and future works are

described.

2 SLOW INTELLIGENCE

SYSTEM

In this section we introduce and develop a general

framework named Slow Intelligence (SI) Systems

(Chang, 2010). We view SI systems as general-

purpose systems characterized by being able to

improve their knowledge over time using the

working context, expert knowledge and task

methodologies. A SI System is one that given a

particular task is able to reason and provide an

answer after completing a process of enumeration,

elimination and concentration. Such systems are

able to improve their knowledge over time using

expert knowledge: a SI Systems continuously learns,

searches for new knowledge and shares experience

with other peers. After acquiring new knowledge,

the system may answer the same task in a different

way and with different results. In particular, the

proposed system follows two decision cycles. The

first one, defined as a short decision cycle, provides

an instantaneous response to the environment. The

second one, a long decision cycle, tries to follow the

gradual changes in the environment and analyze the

information acquired by experts and past

experiences. In this way, the long decision cycle can

influence the short one improving the reliability of

the system. Therefore, SI Systems work in synergy

with the environment and are usually correct but not

always fast. A SI System differs from expert systems

in that the learning is not obvious. A SI System

seems to be a slow learner because it analyzes the

environmental changes, and carefully absorbs that

into its knowledge base maintaining synergy with

the environment. In general, a SI System acts

according to five main phases:

• Enumeration: in this phase a SI System

enumerates all the possible methodologies for

the resolution of a task

• Adaptation: in this phase a SI System acquires

information on the context where it is working

• Elimination: In this phase, a SI System,

according to the information acquired in the

previous phases, selects the best methodology

to approach and solve a task. Information

acquired from experts as well as learned

experiences are used.

• Concentration: After the selection of the best

methodology for solving a task, a SI System

concentrates its resources in solving the

problem.

• Communication: After the resolution of a task,

a SI System updates its experience and shares

the new information with other peers.

3 A PRODUCT CONFIGURATOR

BASED ON THE SIS

APPROACH

In this paragraph the design of the ontological

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

516

configurator, according to the Slow Intelligence

Approach, will be described. First of all some

general definition on the configuration problem will

be introduced.

Definition 1: The configuration problem (CP) is

formulated as: CP = {C, P, C

, R} where C is a set of

components that may constitute a customizable

product, P is a set of properties of components, C

a set of constraints imposed on components due to

technical and economical factors and R is a set of

customer requirements which are usually specified

in the forms of constraints. The CP collects the

user’s request D

Definition 2: The configuration solution (CS) is

defined as: CS = {I, V} where I is a set of

individuals which are instances of components, V a

set of values which are assigned to properties of

individuals. The CS is expressed by the description

of the customized product D

Definition 3: The ontology of the functionalities

) defined as {C

, A

, H

, R

, Ax

} where C

is the concept set.

Cc ∈ expresses one concept

and in each ontology there is ever a root concept

marked as “Thing”. The aim of this ontology is the

representation of the product’s functionalities

requested by the user.

Definition 4: The ontology of components (O

)

defined as {C

, A

, H

, R

TCP

, R

} where

is the concept set.

Cc ∈ expresses one

concept and in each ontology there is ever a root

concept marked as “Thing”. The aim of this

ontology is the representation of the product’s

functionalities requested by the user.

Definition 5: The ontology builder module. This

module has the aim to build an ontology starting

from some inputs furnished by users or obtained by

the environment. In particular this module has inside

a meta-ontology representing in a very general way

the expected ontology. The inputs represent the

nodes and the related relations that have to be

considered in this meta-ontology in order to obtain

the desired ontology. If some inputs are not

represented in the meta-ontology the user will define

the new nodes, the attributes and the relations with

the other nodes of the meta-ontology.

Definition 6: The ontologies’ comparison, simplifier

and merging module. This module has the aim to

manipulate the input ontologies in order to obtain as

output an ontology representing them. In particular

the following operations can be accomplished:

• Merging: The merging operation is so defined:

OOxOM →:

This operation has the aim to

merge two ontologies.

• Simplifying: The simplifying operation is so

defined:

OOxOS →:

This operation has

the aim to simplify an ontology erasing nodes,

relations or attributes. This operation involves

the use of another ontology containing the

information that will simplify the first one.

• Comparing: The comparing operation is so

defined:

ROxOC →:

This operation has

the aim to compare two ontologies giving a

positive or negative grade which is function of

the number of similar nodes, relations and

attributes that are between the ontologies.

• Selection: The selection operation is so defined:

OOxOxTSel →: This operation has the

aim to select an ontology according to the

grade

∈

obtained by the use of the

comparing function.

• Deletion Node: The deletion node operation is

so defined:

OOxODN →: The operation

DN(O

, O

) delete all the nodes in O

that are

in O

• Deletion Relation: The deletion relation

operation is so defined:

OOxRRN →: The

operation DN(O

, R

) delete all the relations R

among the node of O

• Set Attribute: The set attribute operation is so

defined:

OOxOSA →: The operation

SA(O

, O

) set the attributes for each node C

which belongs both to O

and O

to the value

expressed in O

4 THE PROPOSED APPROACH

The proposed approach is based on a four-layer

modelling architecture as shown in figure 1.

Figure 1: Four-Layer modelling architecture.

ONTOLOGICAL CONFIGURATOR - A Novel Approach

517

The first layer expresses the requests furnished by

the user. The requests are collected by the use both

of graphic user interfaces both of text areas. The

second layer maps the collected requests in the

functionalities that the product has to show. The

product’s functionalities are inferred by the analysis

of the user’s requests and are expressed by the use of

the ontological formalism. The third layer has the

aim to select the components achieving the

requested functionalities and also in this case for

their representation the ontological formalism is

adopted. The last layer translates the components’

ontology in the final configuration of the product. In

this case the output is an XML file that can be easily

managed in order to build the customized product. In

order to implement the previous layered architecture

the Slow Intelligent Approach is adopted. The

framework is depicted in figure 2.

Figure 2: The System Architecture.

This system is compatible with the Slow Intelligent

approach and in particular the main blocks can be so

identified:

• Enumerator Block: this block has as inputs the

user request and the product definition and as

output the ontology O’

and has as components

two ontology builders and a comparison,

merging and simplifier modules. This block can

be characterized by the following function:

OOOF

→×: At the end of this block the

ontology representing a first rough version of the

customized product is obtained

• Adaptor Block: this block has as inputs the

ontology O’

, representing a first rough version

of the customized product, and the information

about the context and as output the ontology

O’’

and has as components an ontology builder

and a comparison, merging and simplifier

modules. This block can be characterized by the

following function:

OOOF

→×: At the

end of this block the ontology representing a

context adapted product is obtained

• Eliminator Block: this block has as inputs the

ontology O’’

and the previous product’s

ontologies developed both in the past both by

other similar configurators that work in other

part of the system. The aim of this block is the

tuning of the O’’

according to the previous

configurations obtained in similar contexts. This

block can be characterized by the following

function:

OOxOF

→: The output of this

block is the ontology O

which represents the

adapted product.

• Concentrator Block: this block has as inputs the

ontology O

and as output the configuration of

the product. The aim of this block is the mapping

in an XML file of the ontology representing the

customized product. his block can be

characterized by the following function:

CPC

DOF →:

In this way the configuration problem CP can be

formulated in its general formulation as:

, U

)))). As previously said a Slow

Intelligent System, as the proposed configurator, can

follow a slow and a fast process of solution’s

inference. So the previous formulation can be

defined as the slow process, while the fast process

, U

)). In the next paragraph the detailed

description of each previous introduced block will

be explained.

The Enumerator Block: this block has the

following aims:

• To collect the user’s request and transform it in

the ontology O

• To collect the product’s definition and transform

it in the ontology O

• to map the ontology O

in the ontology O

order to obtain the ontology O’

of the user’s

desired product

In particular the user’s request is collected by the

use of graphic user interface or by the introduction

of sentences explaining the functionalities requested

to the product. This information is managed by an

ontology builder that extracts concepts and relations

organizing them in an ontology which expresses the

desired functionalities of the product. In the same

way an ontology representing the product’s

components and their relations is developed. In this

case the input is the general configuration of the

product, the components and their attributes. The

final result is an ontology O

that represents the

general configuration of the product. The module

Comparison Merging Simplifier has the role to

match the functionalities ontology components with

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

518

the components ontology. At the end of this phase

an ontology O’

representing a first answer to the

user request.

The Adaptor Block: this block has the following

aims:

• To collect the environment information and

transform them in the ontology O

• To adapt the ontology O’

in the ontology O’’

by the use of the ontology O

The Eliminator Block: this block has the following

aims:

• To compare the ontology O’’

with the other

ones that are previously obtained and proposed

to the customers. In this phase a comparison with

the results obtained by other similar modules that

work in different scenarios will be conducted

• To define, after the comparisons, the ontology

The Concentrator Block: this block has the

following aims:

• To implement the product defined by the

ontology O

selecting the appropriate

components

In order to show how the system works an example

will be described. In particular the configuration’s

problem of a personal computer will be faced. The

first step is the introduction of the ontology which

defines the functionalities and the components

related to a personal computer. This ontology,

namely O

GCGF

OO ∪ , is defined by experts

and contains both the functionalities both the

components that can realize them. The user request,

as previously said collected by the use of a graphic

user interface, will be transformed in an ontology

. This ontology expresses the requests of the

customer in terms of the main functionalities that the

final product has to show. At this point the

enumerator block works on the previous ontologies

in this way: first of all the simplifying function

S(O

GP,

) is applied. In this way an ontology O

UFC

containing both the functionalities both the needed

components. The ontology O’

is obtained by the

use of the function DN(DR(O

UFC

, “is_realized”),

). In this way the function of the block

Enumerator is the following O’

= F

GP,

) =

DN(DR(S(O

GP,

), “is_realized”), O

). This

ontology is the input of the adaptator block. In this

case the O’

has to be compared with the ontology

representing the information obtained by the

environment O

. This ontology can be obtained or

by the analysis of the environment where the

customer lives or the product will be used or by the

support of experts. The adaptor block works in the

following way: first of all the function merging is

invoked: M(O’

AP,

). The ontology O’’

obtained by the use of SA fuction: O’’

SA(M(O’

AP,

), O

). In this way the context

adaptation of the ontology representing the product’s

components is obtained. So the function F

is so

defined: O’’

= F

(O’

AP,

) = SA(M(O’

AP,

). The ontology O’’

is the input of the

Eliminator Block which has the aim to compare the

obtained components’ ontology with the other ones

previously developed by other systems that are

working in the distributed system. In particular the

selection function will be used: a threshold will be

fixed and

∈

∀

Previous Ontology Set O

Sel(O’’

, O

). So the function F

is so defined: O

= F

(O’’

, Set_Previous_Ontology) = Sel(O’’

)

∈

∀

Set_Previous_Ontology. At this point

the obtained ontology will be elaborated by the

concentrator in order to obtain the description, in a

XML format, of the components’ list needed for the

realization of the product. So the function F

is a

parser from the OWL language to the XML file

required for the assembly of the product: D

). In order to better explain the proposed

methodology an example will be introduced. A

customer would like to buy a Personal Computer in

order to play videogames and surf on internet. He

knows that he needs an operative system, a browser

web and an antivirus. In particular he prefers a

Microsoft Windows family operative system. He is

in the United States and prefers to have a desktop.

He prefers, besides, cheaper components. This user

by the use of a graphic user interface can easily

define the ontology O

that contains the

functionalities the he needs. Starting from the

general ontology O

containing all the

functionalities and the components involved in a

personal computer. So these ontologies, O

and

, can be used in the function F

obtaining the

ontology O’

.At this point the adaptation module

has to be used. In particular as previously said this

module aims both to introduce new concepts both to

set the attributes of the O’

according to the

information obtained by the environment, where the

product will work, or experts. In the proposed case,

for example, in the proposed configuration there is

the node “keyboard” with the following attributes:

US layout, Euro layout, CN layout: in this case the

attribute US layout will be selected and the other one

will be deleted. So after this phase the ontology

O’’

is obtained and the eliminator block can work.

ONTOLOGICAL CONFIGURATOR - A Novel Approach

519

Figure 3: Particular of the OAP ontology related to the component keyboard.

As previously said this module has the aim, by the

use of selector function, to compare the ontology

O’’

with the other ones developed by other

configurators that work in the system or by the same

configurator in the past. If for example a previous

developed configuration has the same configuration

of O’’

but introduces a node representing a

component, for example a particular kind of

peripheral as a Joystick, for the improvement in the

video game playing experience, it will be the new

ontology O

. At the end of this module the

concentrator block will define the real configuration

by the selection of the real components that are

leaves in the ontology O

(fig. 4). Each node,

representing a real component, contains attributes on

the main characteristics of the products (price,

colour, connection, …)

The selection task will be accomplished according to

the information furnished by the user (i.e. select the

cheaper components), by the comparison of previous

configuration furnished in the past and by the rules

codified in the ontology (i.e. an Intel processor

needs an Intel motherboard). So at the end the

process an xml file containing the final configuration

of the personal computer. In this case this file will be

showed to the customer by the use of a browser web,

so in this way he can check the correctness of the

configuration. The same XML file can be used for

the real configuration of the product.

5 CONCLUSIONS

In this paper we introduced a Slow Intelligent

System approach for the design of a product

configurator based on the ontological formalism. In

particular first designs of the framework and a first

prototype have been developed. In the future a more

detailed experimentation of the system will be

developed.

REFERENCES

A. Felfernig, G. Friederich, D. Jannach, M. Zanker,

“Semantic Configuration Web Services in the

CAWICOMS Project”, 6th IEEE International

Symposium on Wearable Computers, ISWC 2002,

LNCS 2342, pp. 192-205, 2002

X. Ding, “Product Configuration on the Semantic Web

Using Multi-Agent”, Networking, Sensing and

Control, 2008. ICNSC 2008. IEEE International

Conference on, pp. 304 – 309, 2008

N. Franke, F. T. Piller, “Configuration Toolkits for Mass

Customization: Setting a Research Agenda”,

International Journal of Entrepeneurship and

Innovation Management, No. 1, 2003

J.M. Park, J.H.Nam, Q.P. Hu, H.W. Suh, “Product

Ontology Construction from Engineering Documents”,

Smart Manufacturing Application, 2008. ICSMA 2008.

International Conference on, pp 305 – 310, 2008

H. Youliang, Ng Wee Keong, L. Haifeng, L. Wenfeng, S.

Bin, L. Xiang, “Semantic Modeling and Extraction for

Cross-family Product Configuration, Service Systems

and Service Management, 2007 International

Conference on, pp 1-6, 2007

Dong Yang , Ming Donga, Rui Miaoa, “Development of a

product configuration system with an ontology-based

approach”, Computer-Aided Design, Volume 40, Issue

8, August 2008, Pages 863-878

S.K. Chang, "A general framework for slow intelligence

system", International Journal of Software

Engineering and Knowledge Engineering, 20-1, 2010

ICEIS 2010 - 12th International Conference on Enterprise Information Systems

520