A KNOWLEDGE SERVER FOR SUSTAINABLE AGRICULTURE

Main Computing Features

Vincent Soulignac

, Eva Lambert

, Catherine Roussey

, Jean-Pierre Chanet

Jean-Louis Ermine

, Jean-Luc Paris

and Olivier Devise

UR TSCF, Irstea, Aubière, France

Ecole de Management, Evry, France

CNRS, UMR 6158, Limos, Aubière, France

Keywords: Sustainable Agriculture, Knowledge, Information System.

Abstract: Agriculture must evolve into a more environmentally-friendly approach, while remaining economically

viable and socially interesting, which is necessary so that the process can be pursued in the long term, i.e

that the process is sustainable (Brundtland, 1987). Sustainable agriculture has a systemic logic and therefore

requires a strong knowledge base. In this study, we propose a knowledge management computing tool. In

the first part of our article, we discuss the potential actors of the tool and their possible implications. The

second part deals with its contents. In the third part, the main computing features of the tool are shown.

1 INTRODUCTION

Agriculture is involved in a vast societal movement,

imposed on it by the framework and the values

associated with sustainable development. To make a

success of this transformation, agriculture will have

to become both integrated into its environment and

organic. This transformation depends largely on the

mobilization of human knowledge (all that is held

for known or known by an individual or a given

society) and know-how (Ikerd, 1993; Cerf, Gibbon

et al., 2000). But in 2011, while numerous

professional software packages are accessible to

farmers, no structured, interactive IT tool for

knowledge management is available to them. We

thus suggest developing a knowledge management

tool dedicated to organic farmers, called KOFIS:

Knowledge for Organic Farming and its Innovation

System. In the first part of our article, we study who

the actors of this tool are, and their possible roles.

The second part deals with the contents of the tool.

In the third part, the main computing features of

KOFIS are shown.



2 ROLE OF THE ACTORS IN

KOFIS

Not all the actors have the same importance.

However farmers, researchers, agricultural advisers

have not the same expectations, the same needs. The

actors have also to share the same objectives. We

distinguish various accesses to KOFIS according to

the type of actors.

Agricultural advisers or agricultural teachers can

monitor and transfer academic knowledge stemming

from research. Rooted in the institutional

environment, external actors such as cooperatives

influence farmers by their specific requirements, and

this implies a knowledge adaptations process. Their

intervention will also be important via their

contributions on innovative subjects in more open

exchange spaces. The experience and creativity of

agronomic engineers from cooperatives and traders

can be very useful here. The exchange and

innovation will be opened to all the actors to

encourage the new ideas.

3 CONTENTS OF KOFIS

We have decided to concentrate our study on arable

farming. In 2008, 19% of organic farms were

specialized in this type of agriculture.

3.1 Main Content Elements

We have noticed that most of the knowledge about

525

Soulignac V., Lambert E., Roussey C., Chanet J., Ermine J., Paris J. and Devise O..

A KNOWLEDGE SERVER FOR SUSTAINABLE AGRICULTURE - Main Computing Features.

DOI: 10.5220/0003896005250530

In Proceedings of the 8th International Conference on Web Information Systems and Technologies (WEBIST-2012), pages 525-530

ISBN: 978-989-8565-08-2

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

sustainable agriculture system are not generalizable

on a large scale and is, rather, contextualized.

We wish to obtain cognitive representations of

critical knowledge, particularly in order to design

Innovative Agriculture Systems (IAS) which are

effective and sustainable in their context. To do this,

we distinguish two types of mobilizable cognitive

resource: thematic knowledge and contextual

knowledge.

3.1.1 Thematic Knowledge

Thematic knowledge is agronomic, economic or

environmental knowledge. It is generalizable for all

types of farm. It applies only in part to any particular

farm.

3.1.2 Contextual Knowledge

On the farm level, IAS illustrates contextual

knowledge. The notion of reference introduces a

cognitive concept which is specific to agriculture. A

reference is therefore both agricultural advice and

localized facts enabling the data to be interpreted.

References which illustrate the theoretical operation

of a farm could be present in the library, as either a

typical or a concrete case.

A typical case is a “fictitious farm, constructed by

modeling, and described using concrete and coherent

data from farms using the same system” (Cerf and

Lenoir 1987). The typical case is cognitively

efficient for transmitting knowledge which is tried

and tested in a given environment to the operational

actors.

Similarly, a concrete case is "a typical case studied

because of the innovative character of some of its

features, but whose representativity is generally

minor in the local or regional territory. The main

interest of this concrete case is that it can provide

possible orientations, strategies and adaptations of

the main farming systems in the local region.”

(Chambre régionale d'agriculture de Bourgogne

2009).

3.1.3 Which Model to Use?

The tool to be constructed is first of all a

computerized book of knowledge. We selected the

Mask method (Ermine 1996, 2

edition 2000) used

in knowledge engineering to represent both thematic

knowledge and the operations of a farm. These

models are gateways to deeper forms of knowledge,

such as texts, and also perhaps images and videos

(Moity-Maïzi and Bouche 2008). Thus, the models

structure the knowledge.

3.2 Content–actor Relationship

Agricultural advisors or farmers have a role to play

in renewing knowledge, since they are continually

introducing new practices. Researchers, for their

part, make available academic knowledge.

Agricultural advisors or teachers comprise the

pedagogical interface of this empirical and academic

knowledge.

Finally, in the exchange and innovation space, all

the actors can exchange their ideas concerning

innovative subjects, or can react to knowledge which

has already been presented.

4 THE TECHNOLOGICAL

COMPONENT

We will present the general specifications of KOFIS

and then, based on a list of KOFIS properties, we

will propose the general tool architecture.

4.1 KOFIS – General Specifications

In the four following sections, we will describe

successively what is globally required of the tool.

4.1.1 What are the General Expectations for

KOFIS?

KOFIS needs to be accessible to actors of different

origins in various computing environments. In the

first two sections, we identified two expression

spaces with different logical approaches: a

knowledge capitalization space and an innovation

space. In the following, by convention, we will call

[K] the knowledge space and [I] the innovation

space. Thus, KOFIS stores operational knowledge

objects in [K] and builds interactions in [I] to

produce new knowledge.

One of the key success factors for knowledge

management (Soulignac, Chanet et al. 2009) is the

users’ appropriation of the tool. The first property

of the tool is its ergonomy.

Knowledge objects should be quick to access. In

addition, when faced with an unsolved problem, the

system rapidly provides pertinent and explicit

knowledge in the form of datasheets, images or

videos, which facilitate the solution.

Two important functional requirements emerge: the

knowledge capitalization and the organization and

structuring of explicit and tacit knowledge. We shall

see how an adaptation of the C-K theory of

(Hatchuel and Weil 1999) and the use of a semantic

WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies

526

web respond respectively to these two requirements.

4.1.2 Proposed Responses: Capitalization

and Innovation

The C-K design theory (Hatchuel and Weil 1999)

distinguishes two spaces, one associated with tried

and tested knowledge, the other devoted to

innovation, and it reflects human reasoning when

faced with a problem.

Like the C-K theory, KOFIS therefore has two

different spaces: an exchange space [I] where all the

actors can discuss the resolution of a problem, and a

knowledge capitalization space [K] where the

knowledge is stored in a structured way. From the

space [I] it is possible to query the closed space [K].

The discussions of the exchange space allow

enriching the knowledge of the capitalization space.

A second property of the tool is its capacity to

maintain a trace of human design reasoning.

The activity diagram in figure 1 shows the sequence

of actions and decisions within an activity of a

member of <agricultural knowledge system>.

Figure 1: Main KOFIS activities for an actor of the

agricultural knowledge system.

The actors of the <agricultural knowledge

system> are responsible for maintaining a coherent

tree structure for the innovative solutions. They also

validate knowledge coming from this innovation

space in order to publish it in the [K] space.

4.1.3 Proposed Responses: Structuring the

Knowledge

Most web tools have no separation between

Knowledge and data presentation language; only

human intervention can retrieve and process

knowledge. One solution is to annotate the

documents.

Social web-type annotation is based on tagging

documents. A tag is a lexical marker which is

associated with a resource. The search tool will thus

retrieve all the resources associated with the selected

tag.

A second type of annotation, associated with the

semantic web, facilitates the use of knowledge by

machines. The semantic web initiative is supported

by W3C, the international consortium which

standardizes web technologies.

The different semantic web technologies: URI

(Universal Resource Identifier) addresses locate the

resources. The XML language proposes a syntax to

structure all types of contents. RDF and RDFS

represent knowledge in triplets of resources of the

form <subject, predicate, object>. They describe

classes and properties. These triplets are the basis of

the formal description language OWL used to

describe ontology. This language has mathematical

bases, which is why OWL can be manipulated by

computers. In addition, OWL is capable of

instantiating classes, which increases its power to

describe reality. Finally, an ontology can be a

reference, which facilitates exchanges between

computers. SPARQL is a tool for querying recorded

knowledge elements. The semantic web as described

by (Berners-Lee, Hendler et al. 2001) proposes a

proof system. Encrypted blocks ensure the reliability

of the sources.

Figure 2: The multiple layers of the semantic web.

(Bernhard Haslhofer, Linked Data Tutorial, 2009).

In hypertext modeling, links are static and their

semantics are not specified. On the other hand, a

semantic annotation, thanks to semantic query,

enables the computer to retrieve the correct

knowledge element directly.

AKNOWLEDGESERVERFORSUSTAINABLEAGRICULTURE-MainComputingFeatures

527

A document annotation system will therefore

facilitate their retrieval, which constitutes the

third property of KOFIS.

We propose that the system should use the

semantic content associated with the identification

of each farmer, with a view to creating the most

appropriate communities with respect to their

centers of interest. This capacity to construct a

pertinent community is our fourth property.

4.1.4 What are the Functional Requirements

of Each User Group?

KOFIS’ users have various different profiles

(farmer, advisor, teacher, researcher…). Thus, the

difficulty is to find a balance between allowing the

largest possible number of actors to publish in the

tool and maintaining control of what is published.

According to their status, the actors access the

contents with varying rights. Thus we find the five

types of actor whose rights are present below in a

hierarchical sequence.

The KOFIS <visitor> has free access; other users

are authenticated by a classic login/password

system. Apart from <visitors>, all users

communicate identity information to the system, in

order to inform the community’s creation process

later on. The ontology, once defined, is relatively

stable; only the <moderator> may change it.

Table 1: Distribution of users types in [I] and [K].

Space type

User type

[I] Innovation [K] Knowledge

Visitor Read-only

Read-only

Institutional

environment

Proposes and

modifies

innovative subject

Posts comments

and annotations on

innovative subject

Agricultural

knowledge

system

Opens, annotates

and modifies new

knowledge pages

Moderator

Moderates

exchanges

Validates

knowledge

Administers ontology

Manages actors

Administrator

Manages architecture

Manages actor types

The analysis of functional requirements gives

rise to a fifth and a sixth property:

- The fifth resides in KOFIS’ capacity to

generate different user types.

- The sixth is its approach to collaborative

publishing and exchange.

We will add a seventh property: the capacity to

conserve a trace of modifications, which enables a

clean version to be reconstructed in the case of error

or damage.

4.2 Architecture of the KOFIS Tool

In the previous paragraph concerning the KOFIS

tool’s specifications, we identified seven main

properties of the tool summarized in table 2 below:

Table 2: KOFIS properties.

Property Description

1 Ergonomy

Trace the human reasoning associated with

the design

3 Semantic annotation

4 Construction of a pertinent community

5 Differentiated management of user types

6 Collaborative publishing and exchange

7 Traceability

4.2.1 Exploration of Different Types of Tool

First, we explored the software market (Balmisse

2006) corresponding to our need for a collaborative

and semantic knowledge management tool:

Table 3: Comparison of technical solutions.

Tool

Properties

CMS WIKI

Semantic

Web

solution

Ergonomy Yes

No, in

particular for

discussion of

knowledge

pages

Yes

Traceability of the

human reasoning

associated with

the design

Sometimes No Yes

Semantic

annotation

Sometimes Sometimes

Yes

More advanced for Wiki than

for CMS

Construction of a

pertinent

community

No Yes Yes

Differentiated

management of

user types

Yes Yes Yes

Collaborative

publishing and

exchange

Wiki

extension

Yes (all the

site)

Yes

Traceability No Yes Yes

WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies

528

CMS - Content Management System – is an IT

application which designs the contents of a web site

and manages its updates dynamically.

A wiki has a strong collaborative aspect for

elaborating documents. This co-construction leads to

a rich knowledge base, the result of combining the

knowledge and the experience of users/contributors

in a particular domain.

The choice of several pertinent frameworks also

enables the complete development of the

application, but at the cost of a considerable

development effort. We compare the three possible

options in table 3.

To produce KOFIS, we therefore chose the

following tools:

The innovation space [I] is devoted to

collaborative reflection about a new subject. We

chose the CMS tool, because of its capacity to

organize content and its popular publication system:

blogs. However, content management systems are

less developed on the semantic level.

Due to its collaborative production capacity and

its semantic dimension, we therefore chose the

semantic wiki tool for the [K] space. However, it

appears complicated to differentiate between the

user profile of these two communities in a wiki tool,

typically very liberal in terms of rights management.

4.2.2 Tool Choice

For [I], we chose Drupal. This is a high-

performance, robust, open-source tool. Drupal

includes blogs, forums, FAQs, etc. and features a

WYSIWYG interface.

Drupal proposes a tree structure “book” system

to organize its contents. For an innovative subject,

this very visual tool can represent research work as

illustrated in figure 3. The tree structure illustrates

the C-K theory. Drupal offers a classification system

based on taxonomy. This provides the vocabulary

necessary to tag each blog entry serving to support a

solution. The [I] part has a web 2 type annotation,

and thus does not yet have a semantic web

dimension.

For [K], we require a semantic wiki which

combines the collaborative features of a wiki with

the resources of the semantic web (Ludwig,

O'Sullivan et al. 2004; Khun 2008; Schaffert, Eder et

al. 2009). We chose the SMW (Semantic

MediaWiki) module (Völkel, Krötzsch et al. 2006).

SMW accepts OWL and a semantic query

language, ASK. Each page (knowledge page or user

page) of SMW can be annotated by a semantic

annotation. At this stage, we can notice that the [I]

and [K] parts thus have no same type of annotation.

However these two annotation systems are

hierarchically organized in the form of taxonomies;

we have thus worked on taxonomy alignment

(Chhuo Vanna, 2011). In addition, the social

network thus created would be constructed from

ontology. Thus the system will create different

communities depending on the annotation mode. A

community created from a category will be different

if we associate a property with this category. The IT

architecture is presented in the figure below.

Figure 3: KOFIS architecture.

5 CONCLUSIONS

In the previous paragraphs we described the

specifications and the properties of KOFIS. For

every specification of the tool, for every property we

present a technological solution. So, we noticed that

Web 2, also called the social web and Web 3, also

called semantic web are a good solution for a

knowledge server. KOFIS architecture exploits the

capacities of web 2, as well as the latest

developments from the semantic web. Thus the tool

enables the collaborative construction and storage of

knowledge in the [K] space and exchanges in the [I]

space.

Although Web 2 combined in Web 3, answers

our need. Both technologies do not collaborate still

in a satisfactory way. That is why, we work about

the taxonomy alignment, and knowledge querying of

[I] towards [K].

REFERENCES

Balmisse, G. (2006). Outil du KM Panorama, choix et

AKNOWLEDGESERVERFORSUSTAINABLEAGRICULTURE-MainComputingFeatures

529

mise en oeuvre Seconde édition actualisée, Knowledge

consult: 81 pages.

Berners-Lee, T., J. Hendler and L. Ora (2001). The

semantic Web. Scientic American: Volume 284(5)

Pages: 34-43.

Brundtland (1987). Our Common Future

Cerf, M., D. Gibbon, B. Hubert, R. Ison, J. Jiggings, M.

Paine, J. Proost and N. Röling (2000). Cow up a tree

Knowing and learning for change in agriculture, Case

studies from industrialised countries. Paris, INRA.

Cerf, M. and D. Lenoir (1987). Le développement agricole

en France, Presses universitaires de France Que sais-

je?

Chhuo Vanna (2011). Modélisation sémantique d'un outil

de gestion des connaissances pour une agriculture

durable. Rapport de Stage de Master recherche 2

ème

année, Clermont-Ferrand, CEMAGREF

Ermine, J.-L. (1996, 2

ième

édition 2000). Les systèmes de

connaissances. Paris, Editions Hermès.

Haslhofer Bernhard. (2009)Linked Data Tutorial

Hatchuel, A. and B. Weil (1999). Design-oriented

organisations, toward a unified theory of design

activities. 6th International Product Development

Management Conférence, Cambridge, UK, Churchill

College pages 1-28.

Ikerd, J. E. (1993). "The need for a systems approach to

sustainable agriculture." Agriculture, Ecosystems and

Environment, 46: Pages 147-160.

Khun, T. (2008). AceWiki: Collaborative Ontology

Management in Controlled Natural Language. The 3rd

Semantic Wiki Workshop, Tenerife.

Ludwig, L., D. O'Sullivan and X. Zhou (2004). Artificial

Memory Prototype for Personal Semantic

Subdocument Knowledge Management. 3rd

International Semantic Web Conference, Hiroshima.

Moity-Maïzi, P. and R. Bouche (2008). Researchers and

collective know-how: looking at knowledge

production through a lens. 8th European IFSA

Symposium "Empowerment of the rural actors: A

renewal of the farming systems perspective",

Clermont-Ferrand (France), IFSA.

Schaffert, S., J. Eder, S. Grünwald, T. Kurz and M.

Radulescu (2009). KiWi – A Platform for Semantic

Social Software (Demonstration). The Semantic Web:

Research and Applications, Springer Berlin /

Heidelberg 5554: 888-892.

Soulignac, V., J.-P. Chanet, D. O. Paris Jean-Luc and N.

Gondran (2009). Knowledge management and

innovative design, state of the art. 11th International

Conference on the Modern Information Technology in

the Innovation Processes of the industrial enterprises,

Bergamo, Italy.

Soulignac, V., J.-L. Ermine, J.-L. Paris, O. Devise and J.-

P. Chanet (2011). Un serveur de connaissance pour

l'agriculture durable. 4

ème

conférence francophone

GEstion des Connaissances, Société et Organisations

(GECSO), Clermont-Ferrand.

Völkel, M., M. Krötzsch, D. Vrandecic, H. Haller and R.

Studer (2006). Semantic Wikipedia. Proceedings of

the 15th international conference on World Wide Web.

Edinburgh, Scotland, ACM.

WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies

530