article_Sidhom_Lambert__SIIE2011_v5F_EN__ss

Information Design for “Weak Signal” detection

and processing in Economic Intelligence: A case

study on Health resources

Sahbi Sidhom * and Philippe Lambert * *

* Loria/Kiwi & Nancy Université, 4 Rue Ravinelle, 54000 Nancy, France

sahbi.sidhom@loria.fr

** VinaLor, Nancy Université, 4 Rue Ravinelle, 54000 Nancy, France

philippe.lambert@vinalor.fr

Received 20 February 2011; received in revised form 22 November 2011; accepted 25 December 2011

Abstract: The topics of this research cover all phases of “Information Design” applied to detect and profit

from weak signals in economic intelligence (EI) or business intelligence (BI). The field of the information

design (ID) applies to the process of translating complex, unorganized or unstructured data into valuable and

meaningful information. ID practice requires an interdisciplinary approach, which combines skills in graphic

design (writing, analysis processing and editing) , human performances technology and human factors.

Applied in the context of information system, it allows end-users to easily detect implicit topics known as

“weak signals” (WS). In our approach to implement the ID, the processes cover the development of a

knowledge management (KM) process in the context of EI. A case study concerning information monitoring

health resources is presented using ID processes to outline weak signals. Both French and American

bibliographic d a t ab a s e s were applied to make the connection to multilingual concepts in the health watch

process.

Keyword: Economic Intelligence, Business Intelligence, Information Design, Weak Signals

Available for free online at https://ojs.hh.se/

Journal of Intelligence Studies in Business 1 (2011) 40-48

mailto:sidhom@loria.fr
mailto:lambert@vinalor.fr
https://ojs.hh.se/

1. Introduction

On November 2 6 t h . (2010), the Uni ver si t y o f

Califo r nia officially launched its laboratory

project on “Information Design” (ID). The project

aims to develop knowledge exchange between

different actors through applications for new

media platforms such as Ipads in networks or

Iphone Technology. Beyond the innovative

aspect of this project, we note that the ID is the

projection of an important “Prospective

Approach” in the Anglo-Saxon research world.

This point is reinforced by the comparison of the

scientific literature on the issue. Since the 70s,

research teams have specialized on the

connections’ between the graphical representation

of information and its interpretation. One of the

representation techniques that have been developed

is “spatial” information across neuron networks.

Especially in France, this approach has been

somewhat vulgarized at first as in the example of

mind-maps a nd Mind Mapping in education

research. In recent years, this research focus has

been applied to data mining u si n g d ata from the

Web (Web Mining). It helps to develop new

knowledge from large text themes. This technique

is increasingly interested in leaders who have the

responsibility to detect topics that can have been

missed in a linear reading. In the field of EI

studies, the implicit properties on analysis take

on the name “weak signals” (WS) (the explicit

properties are “strong signals”).The detection of

WS allows the user to take better account of the

environment in a dynamic sense and t o b u i l d

foresight (“To prepare today for tomorrow”).

However, the connection between ID and WS

detection requires the development of a

complex methodological process. This is the topic

of this paper. The first part of the paper defines the

meaning of weak signals and processes through a

strategic approach. The second part presents the

logic in ID processes that tends to present varied

graph data sets while facilitating the appropriation

of “semantic” properties. The last part is matched

to a case study on “strategic” health watch for

which we use mapping and the visualizing of

information. The study was able to detect a

n u m b e r o f weak signals on scientific and

technical information with the assistance of ID.

2. Mapping information for detecting weak

signals (WS)

Anticipating strategic failures is one of the most

common issues in EI studies. Market volatility,

uncertainties about property p r i c e s and

economic change are signals that announce

future crisis and breaks from crisis. These breaks

may be opportunities or threats in a changing

world of economics where the faculty of

anticipation becomes a powerful strategic

advantage for companies. In 1970, Ansoff

discussed the concept of WS in his first paper

on the subject, entitled “Managing Strategic

Surprise by Response t o W e a k S i g n a l s ” (Ansoff,

1975). He considers the W S as corollary of

organizational factors in the company, especially

due to environmental turbulence a s compared to

the formulation of corporate strategy. In a

following paper he specified the nature of WS,

by defining it as “a warning (external or internal),

events and developments that are still too

incomplete to allow for an accurate estimate of

their impact and/or to determine a full adapted

response” (Ansoff, 1985). In what follows, we s e t

o u t t o determine the theoretical framework and

application of WS.

2.1 Theoretical framework: weak signals (WS)

Any company can engage in a strategic process.

The specificity of the WS lies also in its

potentiality. If one considers the famous S-curve,

w h i c h describes the four phases of the product’s

life (birth, growth, maturity and decline), we can

imagine that WS is a precursor of a new trend

upstream of the cycle. Hence, the importance of

WS detection in a logic of competitiveness. The

term “signal” is ambiguous. If one refers to the

definition in the Treasury of the French language

(ATILF) for the word “signal” we f ind: [in

French] ”Signe convenu par lequel quelq’un Donne

une information un avertissement á quelqu’un le

moment de faire quelque chose”. Ansoff's: [in

English]: A sign by “proactive” value: to capture

WS by the decision-maker via the channel of

intuition (i.e. spontaneous knowledge of the

environment) to cause a request for additional

information (i.e. explicit formulations) from these

signals. Another contribution to the question of

weak s i g n a l s was made by Coffman who has

worked on various aspects of the problem. For

him, a WS is defined as (Coffman, 1997): An idea

that affects the way we trade and the environment

in which we work; a novelty and a surprise in

terms of receiving signals, a noise and other

signals, sometimes difficult to detect among noise

and other signals, an opportunity or a threat to the

organization, often mad e f u n of b y the

"knowledge h o l d e r s ” or experts, w e a k signal

with a substantial period of time before it

matures and becomes a strong signal, therefore,

this signal represents an opportunity to learn,

grow and evolve. Coffman ( 1 9 97 ) also said that

the WS could be of three types: supra-perceptual

signal, perceptible s ig n a l but not recognized

by our mental models, and recognized signal by

our mental models and by which our change in

behavior. In F r a n c e , H. Lesca (2001) propose s a

list of characteristics that define a WS, w h i c h

is close to that of Ansoff. A signal can be

classified as WS if it is fragmentary, embedded in

a mass of useless information (or noise), an

apparent weak and ambiguous meanings, could

not be seen, an apparent low usability, and low

"palpability". In synthesis of these definitions and

presentations, we can consider that a weak

signal is characterized by: a temporal

discontinuity of its discovery, but also by the fact

that it causes a shift (or breaking) in the facts

found by the receiver to arouse/create

measurable interest in the future. The researchers

found, in the notion of breaking or the

“discontinuity”, the reason for the information

flow and design that someone provides

information, a warning to someone; someone

tells i t ’ s the time to do something. Precisely

the opposite that constitutes the “weak signal” in

a strategic watch process. The transmitter of the

information detected as WS do not expect the risk

that competitors become aware of the potentially

innovative nature of the information given. The

adjective term “weak” is also a sematic problem.

The “weakness” of the signal is opposite to the

potential of information designated b y this t e r m .

T h e ter m “weak signal” is defined by “a high

potential for new innovations”. Due to this we

propose to transform the term “sign” above as in

“weak signal” to “Latent Warning Sign” (LWS)

for designated information in a strategic context.

2.2 Application framework: Knowledge discovery

For nearly a decade, several teams of

researchers across the Atlantic have focused on

the subject of Information Design (ID). The

concept, however, remained forgotten in France

until recently. Visualizing information has

g r e a t p o t e n t i a l advantages. Eppler and

Burkhard (2007) gives six main reasons for why it

is important to give priority to this area: it

motivates the receiver, presents new perspectives,

develops memory, encourages the learning

process, captures the attention of the receiver, and

allows structuring and coordinating of

communication. Many definitions present ID as

an art, the art to direct information to create

meaning. Graphic productions goes together with

significant creativity, with formatted, colorful,

animated and multiform information. In addition

to the purely aesthetic s i d e o f this approach,

ID contains intrinsically a new way of thinking

about information and could be summarized as

Karabeg (2002) did in “a new approach to

information”: He explains what ID is by

p r o p o s i n g t h e i m a g e o f a b u s equipped with

“candle flags” (Atilf, 2010). The bus represents a

“modern culture” while the candles symbolize

“traditional” information. We observe what the

author means by this incongruence (dysfunction).

However it can be surpassed by the use of ID.

Moreover, “modern culture” is producing and

consuming information on a lar ge -scale. I D

through the development o f information

technologies can act as a remedy to the problem of

chronic “infobesity”. On this epistemological ID,

more technique is being added, p r o p o s e d b y

men like J o t h a m Fry (2004) in h i s t h e s i s

e n t i t l e d “Computational Information Design”.

The issue of work is to propose a methodology

for data visualization and offer a comprehensive

set of graphical representations to give sense to

implicit relations between connective data.

J o t h a m Fry (2004) presents a classic seven step

process for ID to ensure the transition from data

to knowledge, acquire: acquisition of data from

any medium, parse or split; cutting to provide a

structure of the data and order, filter: filtering to

select only relevant data, m i n e : the sear c h

w h e r e y o u p l a c e the d a t a into a mathematical

context, show: representation where it is

determined, t h a t a simple representation of data

can take, refine: refining to change the simple

representation to more and advanced visual

renderings, and interact: interaction by adding

methods for manipulating data through

visualization.

Besides the purely aesthetic s i d e o f

information (or infographic), ID is at the

crossroads of several fields of scientific

applications. This includes the fields of

visualization techniques to computer graphics for

greater knowledge. W e also s e e the impact I D

h a s in psychology and semiotics. This

d e v e l o p m e n t is to perfect the cognitive and

physiological theories of visual perception and

cultural factors that come into account in the

process of information visualization. Ultimately,

ID enables the end-user, usually an expert whose

skills enable him to interpret the data

represented as graphs, to generate links between

data and knowledge. This knowledge discovery

is not the ultimate goal of the logic of ID. On the

contrary, the new application aims to refocus the

attention of the user to historical data previously

unnoticed. As a result, a new watch cycle will

begin on data previously “unnoticed”. In this case

study, the aim is to illustrate the relation of ID and

Knowledge Data Discovery (KDD). The “Latent

Warning Sign” (LWS) is here a key component of

this new application by t h e emergence of

thematic relations that may improve the

strategic watch process.

3. T h e Case study: the health heterogeneous

resources – project « cronisanté »

During the conference SIIE’2010, there was a

study about reflections on “chronic diseases

management (project 2007)”, by the High

Council of Public Health (HCSP) with INIST-

CNRS i n Fr ance , t o estab lish a n “Information

S y s t e m for Decision Support” powered by a

strategic watch process on the health resources

(Lambert & Sidhom, 2010). The HCSP was

trying to identify how European health systems

manage the problem of “chronic diseases”. The

approach to the problem is based on the WISP

model (i.e. Watcher Information and Search

Problem) developed by P. Kislin (2007). This

model is the extension of a watch approach to

describe the information needs and help the user

(decision-maker and user) to formulate needs. In

the context of this work, the formulation of needs

has been directed towards the bibliographic

references obtained after consulting a business

database (cf. III.D). The strategic issue of this

work, for the user, is to formalize the declarative

rules, by: <<ISSUE>>: = if we do NOT act on the

<OBJECT> and Knowing the state of the

<SIGNAL>, the n the risk is the expected

<HYPOTHESIS>. Where: ISSUE is defined by

an OBJECT of the environment, on which it is

possible to act. A SIGNAL pro mpts t he

decision-maker t o trigger the problem, an

HYPOTHESIS, w h i c h i s t h e r i s k , a s ,

expected consequences, if left unchecked. The

approach in view of the application is to better

target the information needs of the project sponsor,

the HCSP. To achieve this we translate the

strategic issue in a series of dimensions related to

the problem with a set of indicators on t h e

Information Retrieval (IR) process. One can

easily imagine, given the multidisciplinary nature

of the working group and the specific interests of

each expert, that the heterogeneity of the subject

fields represent a problem in the collected

information and for analysis.

3.1 Health resources: Semantic heterogeneity

“Too much infor mation kills information” ha s

become the favorite expression of responsible

users at the time of information o v e r flow . The

ability to q uickly extract relevant information,

while providing added value, creates more

robustness to any surveillance process. The concept

of added value will here be understood as the

annotation process to facilitate access to relevant

information for the user (Sidhom, 2008). Indexing

and reindexing by users (i.e. social tags,

folksonomies, etc.) are in the list of tools for this

process. Furthermore, the quantity of information,

the heterogeneity of resources and information

itself are a problem well known to designers of

information systems. Schematically, we speak of

a dual heterogeneity that is both semantic and

syntactic. Example of this is the syntax for the

heterogeneity of data storage formats (pdf, doc,

xml, etc.), query languages and more generally

across protocol data structure. The semantic

heterogeneity is the differences between the

interpretations of the real world inducing several

terminology uses for the same reality (ontology,

synonymy, etc.) (Goh, Bressan, Madnick and

Siegel, 1999). Later, we will return to this

problem by relying on examples from the

bibliographic databases. The added value gained

from the system use two aspects. The first is the

addition of keywords or comments (i.e. social

tags) by the user to information resources. This

allows customization of information regarding

documentary resources. These anno tatio ns can

feed up the index in a system to improve the

return rate for IR. The second aspect concerns the

“Information Design” process (Ansoff, 1975).

Several studies i n t h e medical sector have

shown that visual information influenced the

decision-making both in a strategic situation (i.e.

care policies) and a therapeutic condition (i.e.

alternatives to hospitalization) (Elting, Martin,

Cantor and Rubenstein, 1999), (Wyatt. 1999). This

logic has not only to refine the conceptual goal

in the system but also to support the iterative

process: – information needs – IR – new

conceptual indicators (proposing). This process

can come from techniq ues o f Knowledge

Management (KM) and Mind Mapping.

3.2 From NLP to Information Visualization

Sidhom (2002) has developed a Morpho-

syntactic Analysis Platform for automatic

indexing and information retrieval (SIMBAD). It

is composed of an Indexing Kernel (i.e. indexing

process) that uses the noun phrases (NP) as

descriptor in NL structures (i.e. to extract

concepts) in text documents (and opens to

multimedia associated to text descriptions). We

use the definition of a noun phrase (NP) as

defined by Le Guern (1989) to place a lexicon

word in the discourse of universe, de facto, this

word i s ejected i n extensional logic, and gives

to NP a repository status, as a reality segment

associated with it. In our context, the NP appears

to be the bearer of a “semantic load”, which makes

it relevant and a central element to bibliographic

information analysis. Around this semantic we

search guides for our analysis on the actual

corpus. Thus, grammar of NP recognition has three

logic levels:

1°/ Intentional level (or natural

language properties), is represented by

the level N. Words are considered free

predicates o r as simple (i.e. the noun

properties) or as complex (i.e. the noun

properties modified by other units:

adjectival units A’ (i.e. A’

A|Adv+A|A+Rel, etc.), expansional

preposition EP (i.e. EP Prep+N’, etc.),

etc.
2°/ Intermediate level (or taking into

account the universe of discourses), it is

represented by the level N’. It i s the

transition from the intentional to the

extensional levels. Words are considered

free predicates with a set construction of

closed predicates to denote objects in the

world (i.e. N’ N+SP|N+A’|…|N)

3°/ Extensional level (or the NP and its

complexity), it is represented by the level

N’’. It is the close operation using a

quantifier that selects a specific element

in the class N of nominal.

These are the existing objects in the world,

referred objects or mind-constructed objects. In

this work, the morpho-syntactic grammar of the

NP has been rewritten for NooJ in two levels:

firstly, the work was to reformat linguistic

resources (dictionaries and grammars) resources

in our possession. During a second time we

developed the finite state transducer of the noun

phrase. Labels existing dictionaries have been

harmonized to match the syntactic graph of NP

(FIG. 1-2).

Figure 1: Syntactic graph of simple np in Nooj

Figure 2: Embedded syntactic graph of complex np

in Nooj

The graph provides numbered phrases

identifying the fitting relations in syntagmatic

level results (Lambert and Sidhom, 2010). In the

logical use of the semantic concepts (i.e. NP and

its properties) from the bibliographic records, the

results on output graphs must be operated by an

end-user. This gives the end user access to pure

information, leaving him free to evolve in the

concepts from a document process: visualizing

information spaces fed by heterogeneous data

sources. This is a support of an economic

intelligence process and for a information design

system (Lesca, Kriaa-Medhoffer and Casagrande,

2009), for the “ChroniSanté” project. In particular,

in information surveillance activities, the process

is a major vector for the emergence of significant

associations after phases of collection, processing

and analysis in a large mass of data and

information. Several solutions to information

mapping software are available. The tool we

used is software under GNU General Public

License (GPL3) called Gephi (http://gephi.org). It

allows the visualization of complex networks.

3.3 Corpus study and indicator valorizations

As part of our core construction, w e m a i n l y

searched bibliographic databases via the multi-

application “Webspir”, a tool that was replaced at

th e start of 2009 by the platform “OvidSP”

(http://www.ovid.com) with features near

equivalent but more robust for users. Three

databases were selected for the constitution of

bibliographic our entities: <Pascal
2
> <PsyInfo

3
>

and <Medline
4
>. The choice to use these three

sources on health information is justified by our

aim to cover as fully as possible the thematic

management on chronic diseases. The basic

advantage of Pascal database is that it presents

European references and includes records from the

databases in public health (BDSP). The Medline

database is centered on U.S. publications, such as

PsycInfo, but with a broader theme in the social

sciences. In synthesis and contrary to this logic

which requires complete topics, our search

equations were developed to deliver results to the

widest possible extent: first, to cover all the sub-

themes on "chronic disease" and, second, to

identify new sub-themes which we had not

originally thought of. The browsing on the three

databases reported: 2097 references to Pascal,

6110 references to Medline, and 2177 references

to PsycInfo. We subsequently refined our search

to select only those published between 2001 and

2009 in French. The result consists of 397

references and 303 references in the duplication

pass. These entities then will be the first synthesis

of our work. The results of the IR process indicate

that "chronic disease" is a new concept in

France, because of the singularity of the model in

the French health system. A second approach has

motivated a second job on the database “Pubmed”

as previously mentioned. The completion of the

“ChroniSanté” project as a “decision support

system” (DSS) or SIAD in French, was faced with

a semantic problem: the rendering of the term

"chronic disease" i s as a concept purely Anglo-

http://gephi.org/
http://www.ovid.com/

Saxon, which brings a series of problems in a

multilingual and “Translation Terminology”. In

fact, the completeness of the study involves a

search process on multilingual literature to define

the best concept and study consisting of what

intersects. It is in this logic that the base

“Pubmed” was viewed with a search for the term

“chronic disease” in the title of the records. The

result is 13,222 records. These had parallel

entities in t h e English language with relation to

the initial multi-base.

4. Connections to latent warning (or weak

signals) in the id process

Applying automatic analysis (NooJ) on our data

entities (ie. as the ID process in phases 1 acquire

2 parse), the complex graph of NP reported 1374

concepts (ie. as the ID phase 3 filter) including

the smallest concepts (ie. the lemma N) to simple

or complex concepts NP (ie. levels N' + N''). The

advantage of this approach is to present to users

the primary concepts (i.e. as the ID phase 4 mine)

in information resources but also secondary

concepts (i.e. the ID phase 5 representation) that

the user does not necessarily think of i n h i s

r e s ea r c h o f i n d i ca to r s : the translation p hase

o f a decision problem into an IR problem in the EI

context. In this case, considering the concept of

“patient” is t h e central tour theme. In practice,

we tend to establish our search for indicators in a

passive acceptation with concepts: “patient

monitoring [(FR) suivi du patient]”, “patient care

[(FR) prise en charge du patient]”, “patient

education [(FR) éducation du patient]”, etc. But

not in an active acceptation, as “patient

involvement [(FR) implication du patient]”, “active

participation of patient [(FR) participation active

du patient]”, etc. (i.e. as the ID phases 6 refine in

iteration). The research of NPs in the titles of

references highlighted ideas that apparently have

no close relation with our themes, but which

nevertheless appear several times in different

references. In this case, on the theme of “cannabis

consumption” it puts a link for “long term

illnesses”. Given these observations, we took for

advantages of select matches, the longest in the

NP. This corresponds to the fitting relation: the

concept o f fitting in (x y), x the longest and

most informative NP; the concept fitted (zw), z

the shortest and the least accurate NP. The

richness of meaning that emerge, allows the

identification of informational collection with

relevant, complex and hierarchical concepts (NP).

These characteristics may go unnoticed in the

linear analysis of an entity. Thus, the process of

the bibliographic entity records on the Platform

NooJ showed satisfactory results based on the

NP and its semantic properties (i.e. the relations of

Tree (T: y x and x z, Fitting (F: x y w) and Belongs

(B: n x / xy)) (Sidhom, 2002). Pascal: Produced

by INIST-CNRS, PASCAL is an international and

multidisciplinary database that identifies literature

in Science, Technology and Medicine.

PsycInfo i s

t h e Database of the American Psychological

Association (APA) and provides access to journal

articles (many are full text), book chapters and

books, research reports and theses and

dissertations in psychology and related fields

(medicine, nursing, sociology, etc.), from the 19
th

century to today.

Medline is a Bibliographic

database produced by the National Library of

Medicine (NLM-USA). It covers all biomedical

fields: biochemistry, biology, clinical medicine,

economics, ethics, dentistry, pharmacology,

psychiatry, public health, toxicology, veterinary

medicine.

Figure 3: Semantic networks based on

bibliographic records: nucleus a n d satellite

connections.

Concerning t h e visualization o f information

(i.e. as the ID phase 7 interacts in iteration),

we tested the application with Gephi

Fruchterman-Rheingold algorithm (Card,

Mackinlay and Shneiderman, 1999) on the results

of extraction with NooJ. This algorithm of

multi-scale force can calculate the force between

two nodes and map complex networks. By its use,

there is much emerging nucleus surrounded by

satellite subsystems that can be considered non-

central themes to the theme target. According to

the analysis of information needs, the user can

focus attention on these satellites nodes to be

considered as “latent warning signs” themes

(Weak signals i n E I ) and give them special

attention (FIG. 3). By NooJ parsing, we present

the gr ap h results of the NP extraction in the

French entity (FIG. 4).

Figure 4: Visualization of nap semantic networks

Based on the analysis of the semantic network,

we observe that the center of the graph (central

nucleus) consists of terms related to the decision-

making analysis: work that we completed in the

process of EI. Thus, for the analyzed entity, it

reprsents terms such as “coverage, care [FR: prise

en charge]” or “chronic diseases [FR: maladies

chroniques]”, new terms like: (“hepatitis C”,

“cardiopathy”, “asthma”, etc.). We notice the

relation that exists between nodes that represent

the semantic connections between terms (FIG. 5).

The usefulness of such a “visual structure”

document for a user, an expert or a decision-

maker, is considerable. It allows presenting an

interactive document likely to bring new

knowledge. In such a semantic logic, it allows

to better understand the complex dimensions

present when concerned with the surveillance or

EI process.

Figure 5: The np core (central nucleus) and

themes.

The visualization of named entities in the data

can also be positioned relative to the documentary

logic. The graph makes it possible to show

relations between concepts (NP) and document

references (FIG. 6). Said differently, we can see

which resources concentrated more concepts and

which ones that is potentially relevant.

Figure 6: Connection between np and information

resources.

The atomic structure for the nucleus concept,

“coverage, care [FR: prise en charge]” is linked

to the bibliographic references. We also note the

secondary concepts like: “management of asthma

[FR: prise en charge de l’asthme]”, “chronic

asthma [FR: asthme chronique] ” and other

concepts which appear in the peripheral area of

the semantic network. This process was applied to

the second data entity in English, with the same

Logic: the extraction of NPs based on graphs

modeling (Silberztein, 2005). Four hundred and

twenty- nine mapped terms are returned. They are

linked with their document resources. This

approach requires of users a “proactive approach”

through the research (mining) graphs proposed in

order to detected new knowledge. In the logic

presented here, it may refine the concepts from

another cultural sphere. In application, the

activation of a term (T) allows us to view notice

records (N1. Ni) in connection with subjects (S1.

Sm) and associated keywords (K1. Kj). The

scenarios for identifying and refining may be

multiple: as an example, from record (Ni) to

keywords (K1. Kj) or vice-versa. The possibility

to link the node "notice” to the source document

by hyperlink feature allows the user to have

access to the document environment in terms of

interest with any subject.

5. Discussion and conclusion

The experience shows that the techniques we

used require automation to achieve a state of

performance, robustness and an acceptable level

of efficiency. The “Information Design” (Ansoff,

1995) process defined as the “art and science of

preparing information so it can be used by humans

with efficiency and effectiveness.” In the ID

process, we evaluated our study areas by the use of

the following aspects: “graph(s)”, “semantic

network (s)”, “project (s)” and “connection(s)”,

to translate them into clear, immediate and

appropriate information for users. In our study, the

user is often the observer, the analyst or expert

and the decision-maker. For us, useful information

is not the increase of information quantity, but on

the contrary, the reduction of it by relevant

information clusters to facilitate its reading and

appropriation ( Lesca, Kriia-Medhaffer and

Casagrande, 2009). This has been discussed and

treated throughout this paper explicitly as the

application o f I D processes in the context of o u r

“chronic diseases” study (project “ChroniSanté”).

For the “Surveillance” process, w e f o u n d

t h a t the information visualized extracted from

the concepts of NPs is useful to actors in a

strategic project in numerous aspects. At first,

information visualization facilitates document

indexing and content of information systems (IS),

information retrieval (IR) systems or decision

support systems (DSS). As an example, for a

bibliographic record or document to be analyzed,

to extract noun phrases in the content w e m a y

convert them into tags. This solution allows any

user of the document to present the key concepts

in the information database (Lesca, Kriia-

Medhaffer and Casagrande, 2009). This may again

encourage a new Logic of “reindexing b y

users”: The user-tags are automatically stored; the

user will add subjective, objective and creative

tags, to give added-value (Harboui, Ghenima and

Sidhom, 2009). Second, the visualization of a

semantic network (based on NPs concepts and

properties) enables the production of new

knowledge. Viewed no des in a semantic

network can indeed be analyzed in a working

group o r tea m to identify new topics related to

business intelligence as convergence and

divergence of represented subjects (i.e. decision-

making needs). This is, to use the “Humbert

Lesca” logic; heuristic processes allow a collective

creation of meaning (Lesca, Kriia-Medhaffer and

Casagrande, 2009). At third, in the surveillance,

information and documentation p rocesses, the

visualization logic and the ID process can bring

out potentially relevant data. On this point, it

should refine results by a statistical analysis: the

use of bibliometric indicators (Salton, Wong and

Yang, 1975) as the TF-IDF (term frequency-

inverse document frequency). For “Economic

Intelligence”, we see that the usefulness of the

ID process goes beyond a simple and literal

translation of IR indicators (i.e. the translation

phase of a decision problem into an IR problem).

For complex and “multilingual” semantic search,

we can take the conceptual differences between

terms, like: “chronic disease” or “chronic disease

management” and “management of chronic

diseases”. Thus, we can show, by semantic

visualizing of these concepts, the connections with

the processed information (parsing, analysis and

needs information). Also, we can establish multi-

level intersections between information concepts

and common or different “morphemes” to get

shared meanings. Finally, on a technical level, the

problem of heterogeneous information resources

can be minimized with the addition of

annotations and/or linguistic processing. NLP

tools (as NooJ or others) will enable the

processing of multi-format sources, and for non-

textual documents (multimedia), analysis can be

made on the annotations associated with the

documents: developing a semantic

homogenization, b y counterbalancing the

syntactic heterogeneity.

References

ATILF – CNRS. 2010. - Université Nancy 2, Le

Trésor de la Langue française,

http://atilf.atilf.fr/dendien/scripts/tlfiv5/advan

ced.exe? 8;s=66859215; [visIted date: Nov.

20, 2010].

H. I. Ansoff. 1975. “Managing s t r a t e g i c

s ur p r i se by response to weak signals”.

California Management Review, vol. 18, no. 2,

pp. 21–33.

H. I. Ansoff. 1985. “Strategic response in

turbulent environments”. Handbook of

business strategy.

S. Card, J. Mackinlay, and B . Shneiderman (Eds).

1999. Readings in Information Visualization:

Using Vision to Think. S.F., CA: Morgan

Kaufmann.

B. Coffman. 1997. Weak Signal Research, Part I:

Introduction, Journal of Transition

Management, MG Taylor Corporation, Jan 1,

2011, from

http://www.mgtaylor.com/mgtaylor/jotm/winter

97/wsrmatur.htm

B. Coffman. 1997. Weak Signal Research, Part II:

Information Theory, Journal of Transition

Management, MG Taylor Corporation,

Retrieved Jan 1, 2011, from

http://www.mgtaylor.com/mgtaylor/jotm/winter

97/wsrmatur.htm

L. S. Elting, C. G. Martin, S. B. Cantor, and E. B.
Rubenstein. 1999. “Influence of data display

formats on physical investigators´ decisions to

stop clinical trials formats on physician

investigators´ decisions to stop clinical trials:

prospective trial with repeated measures,”

BMJ, vol. 318, no. 7197, pp. 1527 -1531.

M. J. Eppler and R. A. Burkhard. 2007.

“Visual representations in knowledge

management: framework and cases”.

Journal of Knowledge Management, vol. 11,

no. 4, pp. 112–122.

http://atilf.atilf.fr/dendien/scripts/tlfiv5/advanced.exe
http://atilf.atilf.fr/dendien/scripts/tlfiv5/advanced.exe

B. Jotham Fry. 2004. Computational information

design. bfa Communication Design, in

Computer Science Carnegie Mellon University.

Massachusetts Institute of Technology, April,

Retrieved 5 Jan from

http://benfry.com/phd/dissertation-050312b-

acrobat.pdf

C. H. Goh, S. Bressan, S. Madnick, and M. Siegel.

1999. “Context interchange: new features and

formalisms for the intelligent integration of

information,” ACM Trans. Inf. Syst., vol. 17,

no. 3, pp. 270-293.

M. Le Guern. 1989. “Sur les relations entre
terminologie et lexique,” Meta, vol. 34, no. 3

A. Harbaoui, M . Ghenima, a n d S . Sidhom.

2009. “Enrichissement des contenus par la

réindexation des usagers: UN é t a t d e l’art sur

la problématiq ue ”. In Proceedings SIIE'09

-Systèmes d'Information ET Intelligence

Economique, IHE Edition. Hammamet (12-

14 fév. 2009).

D. Karabeg. 2002. “Designing information

design”. Information Design Journal, vol. 11,

no. 1, pp. 82–90, 2002.

P. Kislin. 2007. “Modélisation du problème

informationnel du veilleur dans la démarche

d’Intelligence Économique”. Thèse de

doctorat: université de Nancy2. 5 Nov.

P. Lambert, S. Sidhom. 2010. “Knowledge

extraction and visualization: case study on

ChroniSanté project in France “. In Procedings

SIIE'2010, Sousse (Feb. Tunisia). IHE Edition.

H. Lesca. 2001. “Veille s t r a t é g i q u e : passage

d e la notion de signal faible à la notion de

Signe d’alerte précoce,” in Colloque VSST, pp.

98–105.

H. Lesca, S. Kriaa-Medhaffer, and A.

Casagrande. 2009. “Veille stratégique: Un

Facteur d’échec paradoxal largement avér é :

la surinformation causée par l ’ I n t e r n e t . Cas

c o n c r e t s , r eto ur s d ’ e x p é r i e n c e et piste de

solutions”. in Proceedings SIIE'09 - Systèmes

d'Information et Intelligence Economique, IHE

Edition. Hammamet (12-14 fév. 2009).

G. Salton, A. Wong, and C. S. Yang. 1975. “A

vector space model for automatic indexing,”

Communications of the ACM, vol. 18, no. 11,

pp. 613–620.

S. Sidhom. 2002. “Plate-forme d ’ a n a l y s e

morpho-syntaxique pour l'indexation

automatique ET la recherche d'information: de

l’écrit vers la gestion desconnaissances”.

Thèse de doctorat, Université Claude Bernard

Lyon1.

S. Sidhom. 2008. “ Approche conceptuelle par un
processus d'annotation pour la représentation et

la valorisation de contenus informationnels en

intelligence économique (IE)

contenus informationnels en intelligence

é c o n o m i q u e (IE),” In P r o c e e d i n g s

SIIE’08 - Systèmes d'Information et

Intelligence Economique, IHE Edition vol.1,

Hammamet (fév. 2009).

M. Silberztein. 2005. “NooJ: a linguistic

annotation system for corpus processing,” in

Proceedings of HLT/EMNLP on Interactive

Demonstrations, p. 11.

J. Wyatt. 1999. “Same information, different

decisions: format counts,” BMJ, vol. 318, no.

7197, pp. 1501 -1502.

http://benfry.com/phd/dissertation-050312b-acrobat.pdf
http://benfry.com/phd/dissertation-050312b-acrobat.pdf