A Topic-Map-based Framework to Enhance Components’ Retrieval

in a Process Control

Sara Bouzid

1,2

, Corine Cauvet

and Jacques Pinaton

Laboratory for Information Sciences and Systems, Saint-Jerome University, Marseille, France

STMicroelectronics, ZI De Peynier, Rousset, France

Keywords: Components’ Retrieval, Semantic Description, Process Control, User Requirements, Topic Maps.

Abstract: Controlling a manufacturing process to meet customers’ requirements is crucial for many industrial

companies to remain at the forefront. The control of a manufacturing process requires for each actor

involved in this process to be reactive and effective in monitoring the process in order to meet company

objectives. Enhancing the sharing and the retrieval of the components that support the control of such a

process is necessary to effectively meet actors’ requirements for the accomplishment of their business tasks.

Semantic description with knowledge representation techniques is one of the widely used techniques to

address components’ retrieval. Existing methods focus on describing components according to their

functionalities, thus they don’t really tackle user requirements that the components meet. We propose in this

paper a layered conceptual framework to add semantic description to the components that support the

manufacturing process control starting from the users’ requirements. A Topic-Map-based solution is

proposed to support the implementation of this framework.

1 INTRODUCTION

Controlling large scale manufacturing processes in

industrial companies is a complex task that requires

a continuous monitoring and analysis of the process

to ensure its achievement. The manufacturing

process control consists of a set of tools and methods

established to guarantee that products meet

customers’ requirements. To that aim, process

control systems and tools enable to extract many

indicators to follow the manufacturing process and

to measure the adherence to company objectives.

We note that an indicator is an aggregate of

measures (Claudepierre, 2010) as known in the

decision engineering field. The main problem in

establishing such a control activity is giving access

to process control tools to all the actors of the

company that have same needs. The used tools are a

set of heterogeneous components such as a software

product, a software component, a database system or

even an information resource. Thus, sharing and

retrieving such components between users without

an organized way and a consistent method is often

difficult. There is a lack today of an efficient method

to manage the sharing and retrieving of the

components that support process-control methods in

industries. The main difficulty in this context resides

in the distance between what a component does (the

what) and which need it addresses (the why). This

distance can entail low-quality search and results

(Lucredito et al., 2004). The goal of our work is to

reduce this gap in order to improve components’

retrieval during the control of a manufacturing

process. We propose a Topic-Map-based framework

to integrate users’ requirements in the components’

description. The conceptual design of the framework

is based on three levels of abstraction: a requirement

level, a functional level and a physical level. The

characteristic of our proposal is that we use the

Topic-Map standard (Pepper, 2009) which is an ISO

semantic web standard to support the integration of

these levels.

The second section in this paper presents an

overview of the conceptual framework. The third

section presents the Topic-Map-based solution

which supports the implementation of the

framework. The proposed framework with its Topic-

Map implementation is illustrated by an example

from STMicroelectronics.

146

Bouzid S., Cauvet C. and Pinaton J..

A Topic-Map-based Framework to Enhance Components’ Retrieval in a Process Control.

DOI: 10.5220/0003977601460149

In Proceedings of the 14th International Conference on Enterprise Information Systems (ICEIS-2012), pages 146-149

ISBN: 978-989-8565-11-2

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

2 A CONCEPTUAL

FRAMEWORK ALIGNING

COMPONENTS WITH THE

USERS’ REQUIREMENTS

2.1 Overview

Our goal is to define a framework for components’

description to enhance their findability and usability

between users involved in the control of a

manufacturing process. To that aim, we can

distinguish three main types of knowledge

underlying this context: User requirements,

component functions and the components

themselves. We propose then a three layered

framework as follow: Requirements’ level,

Functional level and physical level. Each level is

supported respectively with a meta-model (URM,

FM, PM) to handle the components’ description in

an organized manner.

Figure 1 depicts the proposed layered

framework. The goal of each level is to provide a

semantic description above components driven by

the users’ requirements. In fact, the main difference

between our components’ description and the

existing models proposed by the scientific

community is that our framework is oriented toward

non-expert end-users. Most existing components’

description models are proposed for market

components and are usually devoted to address

retrieval issues in the context of software

development. Our proposed framework is rather

devoted to enhance the retrieval of diverse

components related to a specific industrial process.

Figure 1: Overview of the conceptual framework.

The semantic description proposed in this

framework starts from the users’ requirements and

focus on the relevant properties that can guide the

user in the selection of a given component. In

addition, each level is enriched by concepts from the

manufacturing process ontology because these

components are kinds of services supporting the

control of manufacturing processes. Thus, the

proposed framework enables in fact to align

components with the users’ requirements using

semantic-description techniques.

2.2 The Requirement Level

The requirement level enables to capture the users’

requirements in a manufacturing process control. A

user in this context is an actor that can be a person or

an organization from the process control function.

This user has one or many requirements which can

be simple or complex. Each requirement is then

expressed as a goal that can be refined with an

AND/OR graph (Lamsweerde, 2001). A goal at a

low level is measured with one or many indicators.

These indicators enable at the end to respond to the

user requirement because the actor can know

through an indicator if he reached company

objectives. Figure 2 depicts an example of goal

decomposition until the indicators definition from

STMicroelectronics. STMicroelectronics is a

French-Italian company specialized in the

manufacturing of electronics chips. This example is

related to the Wafer-Fab-Yield (WFY). A Wafer is a

silicon plate used as support for the construction of

chips.

Figure 2: Example of a Requirement Model.

To reduce WFY loss, the company actors have to

follow two main goals: to reduce scraps and to

prevent them. A scrap is a wafer that is not

compliant to the quality and the reliability

requirements. To reduce scraps, process control

actors need to detect scraps and to monitor them. To

detect scraps, three main indicators are supervised:

ATopic-Map-basedFrameworktoEnhanceComponents'RetrievalinaProcessControl

147

the number of accidents, the percentage of lots Out

Of Control (OOC) and the percentage of lots Out Of

Specifications (OOS). These indicators are mainly

based on statistical methods.

2.3 The Functional Level

The functional level describes the what of the

components to enable the user to know what a

component does according to his requirement. In

this level a component is described with its

functionalities and its outputs. The outputs of the

components here are in fact the indicators sought by

the users for the process control. We can add also in

this level some non functional properties. These

properties give information about the quality of

service of the component such as its usability, its

performance or its reliability to help the user in the

selection of a given component.

2.4 The Physical Level

The physical level provides information about the

components themselves as physical entities. It gives

the components with their meta-data to the user. As

kinds of meta-data we can cite the execution

environment of the component, its location and

eventually the used data-sources to generate the

indicators. We note that these components can be

heterogeneous depending on company assets and its

information systems. Some components can be kinds

of decision-support systems - such as reporting tools

and data analysis tools- or artifacts produced with

these systems. Other components can be kinds of

documents and web resources such as dashboards

and web portals referencing process-control

indicators.

We propose to support the implementation of the

framework with the Topic Map standard which is a

semantic web standard for resources’ annotation and

ontologies’ construction.

3 TOPIC MAPS TO SUPPORT

COMPONENTS’ DESCRIPTION

Topic Maps are an ISO semantic web standard

(Arroyo et al., 2004) usually used to build semantic

networks of data and concepts linked to

heterogeneous resources. The key concepts of a

Topic Map consist of topics, associations,

occurrences and resources (figure 3). A topic is a

symbolic representation of a subject where a subject

is a concept from a real world (Pepper, 2009). An

association expresses a relationship between topics.

An occurrence is what links an information resource

to a topic and finally a resource is any technological

support that handles information. It could be a

document, a web page, a software product, a

Database, etc.

Figure 3: Core Concepts of the Topic Map Standard.

We chose to use the Topic Map standard mainly

because it enables to describe heterogeneous

resources with high semantic abstraction. A Topic

Map can therefore represent any subject from the

real world with any desired level of granularity by

typing topics and associations. In addition, Topic

Maps are highly oriented towards human users and

are more optimized for findability (Pepper, 2008)

comparing to OWL and RDF which are more

devoted to machines and interoperability between

applications (Arroyo et al. 2004).

According to our purpose, a Topic Map will

enable to structure the way of navigation and search

of components starting from users’ requirements.

Thus we propose to extend the Topic-Map Meta-

Model as depicted in figure 4 to include the

proposed layered description of the framework. The

added concepts are represented with dark color in

figure 4. The standard concepts in Topic Maps

consist of topics, associations, occurrences and other

characteristics of a topic such as topic name that can

have variant names and association role which

indicates the role of each topic in a binary

association. A resource is also a key concept in a

Topic Map, which can be used to represent any kind

of component. According to our proposed

framework, we can distinguish two types of topics

(figure 4): Requirement Topics and Usage Topics. A

Requirement Topic refers to the user requirement. It

can be a goal or an indicator. A Requirement Topic

can be decomposed and refined into sub-topics

according to the complexity of the user requirement.

A Usage Topic is a kind of topic that describes the

components (resources). A Usage Topic can

represent the component name (CName), its

functionality or its output. The physical level is

represented as meta-properties linked to the

occurrences in the Topic-Map Extended Meta-

model.

ICEIS2012-14thInternationalConferenceonEnterpriseInformationSystems

148

Figure 4: The Topic-Map Extended Meta-Model.

According to these types of topics we distinguish

three layers in the Topic-Map Extended Meta-

Model: the requirement layer (related to requirement

topics), the usage layer (related to usage topics) and

the resources layer. Figure 5 illustrates an example

of a Topic Map exposing these levels. We take again

the example of the WFY goals.

Figure 5: Graphic view of a Topic Map.

The Wandora tool, a free software environment

for Topic Maps creation and visualization, was used

to implement the Topic Map.

Starting from the goal Reduce WFY loss, the

proposed Topic Map enables a structured navigation

between topics so to discover the set of indicators

required for this complex requirement and to obtain

them with the available components of the company.

For example to detect scraps, the Topic Map shows

that the user needs to obtain the %OOC among other

indicators. This indicator is an output provided by

the control card recipe and is accessible via PC

intranet.

7 CONCLUSIONS

This paper tackles how to enhance components’

sharing and retrieval in a manufacturing process

control by reducing the gap between the users’

requirements and the components that support the

process control. The study of the users’ requirements

within STMicroelectronics confirms us our findings

and helps us to propose a suitable framework to

address this need for the control of a standard

manufacturing process. Our current works try to

study in detail the concept of indicator in order to

make a classification of the indicators according to

types of requirements. The linking between the main

levels of the framework can be better specified in

this case. A search application based on Topic Maps

is also planned.

REFERENCES

Arroyo, S., Lara, R., Ding, Y., Stollberg, M., Fensel, D.,

2004. Semantic Web Languages: Strenghs and

Weaknesses. In IADIS04, International Conference in

Applied computing, Lisbon, Portugal.

Claudepierre, B., 2010. Conceptualisation de la

gouvernance des systèmes d’information. PhD report.

Université Paris I-Pantheon Sorbonne.

Lamsweerde, A., 2001. Goal-Oriented Requirements

Engineering: a Guided Tour. In the 5th IEEE

International Symposium on Requirements

Engineering, Toronto.

Lucrédio, D., do Prado, A.F, de Almeida, E.S, 2004. A

survey on Software Components Search and Retrieval.

On Proceedings of the 30

EUROMICRO Conference,

IEEE.

Pepper, S., 2008. Topic Maps and All That: http://topic

maps.wordpress.com/2008/05/11/topic-maps-and-the-

semantic-web/

Pepper, S., 2009. Topic Maps. Encyclopedia of Library

and Information Sciences. Third Edition DOI.

ATopic-Map-basedFrameworktoEnhanceComponents'RetrievalinaProcessControl

149