Creative Colours Specification Based on Knowledge (COLorLEGend ...

3 downloads 0 Views 2MB Size Report
colours specification coming from visual perception, cognitive sciences, graphic ... various colours specifications and maps, depending on cartographic and ...
The Cartographic Journal Vol. 48 No. 2 # The British Cartographic Society 2011

pp. 138–145 International Cartographic Conference, Paris 2011-Special Issue May 2011

REFEREED PAPER

Creative Colours Specification Based on Knowledge (COLorLEGend system) Sidonie Christophe COGIT Laboratory, IGN France, 2/4 Avenue Pasteur, Saint Mande´ 94165, France Email: [email protected]

In map design, user’s choices of graphical signs may be unsuitable to their needs. In particular, colours choices may disturb the reading and understanding of maps. We propose a methodology to help users to make personalized and original colours specifications. It implies offering cartographic expertise and favouring creativity. The paper handles with knowledge for colours specification coming from visual perception, cognitive sciences, graphic semiotics, cartography and art. We specify a knowledge base made-up of cartographic rules (semantic, contrasts and conventional), artistic rules and inspiration sources (map samples and famous paintings). The COLorLEGend system based on this knowledge allows users to make various colours specifications and maps, depending on cartographic and artistic colours uses. Lors de la conception cartographique, les choix des e´le´ments graphiques faits par un utilisateur peuvent ne pas re´pondre a` son besoin. En particulier le choix des couleurs peut perturber la lecture et la compre´hension de la carte. Nous proposons une me´thode pour aider un utilisateur a` de´finir des couleurs personnalise´es et originales. Cela ne´cessite de mettre a` disposition une expertise cartographique et de favoriser la cre´ativite´. Ce papier de´crit des connaissances ne´cessaires a` la spe´cification de couleurs, connaissances emprunte´es aux domaines de la perception visuelle, des sciences cognitives, de la se´miologie graphique, de la cartographie en ge´ne´ral et de l’art graphique. Nous avons de´fini une base de re`gles compose´e de re`gles cartographiques (se´mantiques, contrastes colore´s et re`gles conventionnelles), de re`gles artistiques et de sources d’inspiration (des exemples de cartes et des peintures ce´le`bres). Le syste`me COLorLEGend base´ donc sur ces connaissances permet a` un utilisateur de choisir diffe´rentes spe´cifications de couleurs pour ses cartes selon des usages de couleurs artistiques ou cartographiques. Keywords: map design, colour, knowledge, symbol specification, semiotics, perception

INTRODUCTION

Current cartographic web tools1 and geographic information systems (GIS)2 permit users to quickly make a map with the help of various functionalities: loading and overlaying of geographical and thematic data through online geoportals, points, lines and polygons design, modification of symbol specifications, etc. Some users may have some theoretical and practical knowledge in cartography, but it is not the case of all. Nevertheless, those tools give no substantial help to users to make suitable choices in order to get a final map, adapted to their tastes and needs and legible by future readers: the cartographic message delivered by the map, built in and for a specific context and use, according to specifics needs, may be misunderstood. Anyone can easily become a mapmaker and a ‘mapreader’. Our paper is about cartographic message’s quality, i.e. in its making of but also in its understanding. It depends first on the quality of data and symbol specification chosen DOI: 10.1179/1743277411Y.0000000012

by the mapmaker, but also on supposed cognitive abilities of the map-readers. The term ‘user’ is used here to represent all users of cartographic tools as mapmakers but also as map-readers. In the map design process, the stage of symbol specification is often uncertain and the user’s choices of graphical signs may be unsuitable to their needs. In particular, colours choices, as mistakes or misunderstandings, are often highlighted: too many colours, colours un-adapted to user’s data, too meaningful colours, etc., involving disturbances when readers try to read the map and understand its cartographic message. Our research work handles with knowledge and processes specifically involved in symbol specification and, in particular, in colours choices. The paper presents the results of our PhD thesis concerning the proposition of a methodology to help users to make personalized and original colours choices during map design process (Christophe, 2009). Our purpose is thus to cooperatively build a colours specification with the

139

Creative Colours Specification Based on Knowledge (COLLEG)

user, depending on their tastes and needs. It involves offering cartographic expertise to users and favouring their creativity. In this paper, we focus on the identification, formalisation and management of knowledge involved in this particular step of colours specification. First, we detail why the colours specification process is a complex problem to solve. Second, we assume that it requires identifying knowledge from various thematic domains that we have to handle with. From this identification, we propose a knowledge base for colours specification – cartographic and artistic rules, inspiration sources. In the third section, we present the management of our knowledge base by our COLorLEGend system (Christophe and Ruas, 2009) as constraints on the colours specification in progress.

COLOURS SPECIFICATION PROCESS: A COMPLEX PROBLEM

In the cartographic context, we have to handle with the following semiotic triangle: the map (signifier), the message (signified) and the real objects (referent). We focus on the stage of symbol specification during map design process which is the significant part of the making of the cartographic message. Especially, we consider the colours specification process as a complex problem to solve. Symbol specification process

The symbol specification process is a part of the map design process: the user analyses their needs and selects suitable geographical data; information are put together in geographical themes with the help of relations – association, difference and hierarchy – between them. In a topographic map, various geographic data types – roads, administrative boundaries, buildings, networks, land uses, etc. – are thus grouped by themes: they are represented by graphical signs which differ each other from their geometric type – punctual, linear and surface – and from their shapes, sizes and colours. Graphical signs are then applied on all cartographical objects in the map. The key of a map gives all correspondences between graphical signs used in the map and their meanings. If the final visual representation is not appropriate or satisfactory, the user may modify and improve those signs. Our research work focuses on choices of graphical signs and in particular colours, as a main problem to solve. It aims also at graphically representing the user’s data with the chosen colours and at evaluating and improving the final colours specification. A colours specification of quality is made of suitable objects–colours–meanings associations, easily readable and understandable by any possible readers. Colours specification process

The colours specification process may be described as a set of chosen colours that has to be linked to a set of cartographical objects in the map. It requires first that the set of colours is selected from the whole of possible colours yet. Three aspects are then at stake:

N N

colours have to be adapted to user’s tastes and needs; colours have to be suitable to their related geographical themes: it means that a selected colour for a theme is

N

adapted to the semantic of the theme, facilitating its readability and understanding by another user; colours combinations are suitable to relations: (1) between geographical themes (semantic relationships); and (2) between cartographical objects in the map: it means that selected colours put together in the map, according to shapes and sizes of objects, bring a general feeling, not disturbing but helping the understanding of the cartographic message.

The management of colours has a priori an endless number of solutions: the whole of possible visible colours and their combinations is wide. Among this endless number of solutions, many solutions are potentially acceptable because they fit the problem, but none is perfect or optimal. Only the user can select the solution the most satisfactory for them. We highlight this notion of satisfaction influencing the determination of solutions. According to Simon (1973), three aspects of ill-structured problems can be also highlighted in our context:

N N N

problem-space is not explicitly and exhaustively defined: too many elements could be added, changing constantly the limits of this space; there is no predefined and complete procedure to solve the design problem: the colours specification has effectively no generic solving method in the cartographic theory; the proposed solutions can only be evaluated according to their levels of relative satisfaction: any solution is ‘acceptable and satisfactory, not optimal’.

A non-expert user may face difficulties to build a mental representation of their goal to reach: a map with personalized and original colours choices. To acquire expertise plays a significant role in a better definition of a complex problem. Thus, our proposition of a design method should integrate an expertise to help users to progressively build a mental representation of the goal to reach, during the process. We highlight that the main difficulty in solving a design problem is the level of expertise to acquire. The symbol specification requires actually a lot of non-formalized knowledge coming from various domains. Managing colours requires knowledge not only to manage a colour code, but also feelings, interpretations, tastes and needs.

KNOWLEDGE BASE FOR COLOURS SPECIFICATION

In this section, we identify knowledge required for personalized and original colours specification. This knowledge proceeds from various domains. With the help of this identification, we propose rules of colours uses in cartography. Moreover, we propose inspiration sources as knowledge supports of cartographic and artistic colours uses. Identification of knowledge: visual perception, graphic semiotics, colours theories and colours in cartography

Colour is powerful because of its physical, psychological and cultural aspects involving feelings and interpretations, not necessary shared by everyone. Colour is widely used and abused, first of all on numeric supports, involving mistakes

140 and misunderstandings in visual representations. In order to better manage colours, we have to handle with basic-science cognitive research, graphic semiotics, cartography and painting art. Through this overview, we try to highlight main principles about colours specification in map design process. Perceptive, cognitive and semiotic aspects

Objects in a visual scene can be processed one after the other. This serialisation of visual scene analysis is made through mechanisms of visual attention. We thus focus on visual attention mechanisms allowing selecting a part of the visual information, from the spatial position or visual characteristics of objects. In particular, two mechanisms have been highlighted by Treisman and Gelade (1980) that can be applied to cartography:

N

N

automatic detection, pop-out or visual salience, i.e. what the user sees first and interprets as the most important to see and to remember. In our context of colours specification, a salient colour may be a colour implied in a high-value contrast or in a high-quantity contrast. The salient colour is generally considered to be the colour of the most relevant theme of the map; controlled detection, conjunctive search, depending on more elaborated cognitive functions, as for instance, categorisation: the reader of the map tries to search objects to put together, with the help of visual characteristics which provide sense, which are similar and which are not (Rosch, 1976). In this context, the Gestalt Theory established that a set of objects is perceived as a group, if these objects have similarity, proximity, symmetry, closure or alignment properties. This detection is thus depending on cognitive abilities of users and on knowledge they have. The reading of a map may be processed in three steps: (1) graphical signs given in the key map are mentally represented through their visual characteristics (colour and geographical primitives); (2) cartographical objects in the map are compared to those signs; and (3) cartographical objects are categorized one by one.

In order to make spatial inference making more efficient maps, many researchers currently work on visual perception and basic-science cognitive search (Montello, 2002). Fabrikant et al. (2010) empirically investigate the relationship of thematic relevance and perceptual salience: they ask if perceptually salient elements draw novice viewers’ attention to thematically relevant information, whether or not users have domain knowledge. They thus are interested by the ‘where’ do users look in the map, and not the ‘what’ do they look at, referring to the semantic characteristics requiring more elaborated cognitive processes (Fabrikant, 2005). Bertin (1967) defines seven visual variables – localisation (x,y), size, value, colour hue, shape, orientation and grain – in his book specifying the conceptual framework of the Graphic Semiotics3. He characterizes each visual variable with specific perceptive properties: its capacity to convey relationships existing between data and its length, i.e. the number of level of differentiation the visual variable can give. Graphic semiotics has not been especially described for cartography, but is a main part of cartographic theory.

The Cartographic Journal

Nevertheless, if Bertin’s work helps cartographers to manage a visual variable, it does not describe a generic method allowing managing and combining several visual variables and their related properties, as a whole. Research works about semantic relationships between cartographical objects (association, differentiation and hierarchy) have been driven to propose suitable colours schemes to a specific relationship chosen by a user (Brewer, 2003) or to improve colours in an existing map (Chesneau, 2006; Buard and Ruas, 2009). Those perceptive, cognitive and semiotic aspects help us to take into account specific colours choices involving salient or more cognitive effects, through the use of contrasts and the analysis of several visual variables (not only colour hue or value but also of shape and size of cartographical objects), to render specific semantic relationships. Normative, creative and artistic aspects

We notice that colours are associated with symbolic representations and temperature feelings. According to the cultural context, colours are automatically associated with their physical colours in nature or with the idea people have about them: the sea and hydrographical networks have a blue colour, wooded area have a green or brown colour, and land use has yellow, red or brown colours. About elements which are considered to be physically blue, Mollard-Desfour (1998) categorizes them as ‘all that is naturally blue (in a whole or in part) or that is closed to blue; that seems to be blue or to be of a colour closed to blue’. These associations are called conventional and are understandable by everyone in a cultural and temporal context. Our purpose is to integrate those normative aspects and to face them with more creative ones. To make colours choices is a creative process. It is a fact that creative and artistic abilities allow selecting colours and colours combinations more easily. Conferring capacity in producing visual art is currently studied by neuropsychologists in order to know which part of our brain is involved in creativity (Chatterjee, 2006). In our context, creativity may be favoured if degrees of freedom in map design process are offered to mapmakers: it means that propositions may be done in order, first to improve imagination, inspiration and thus to wider explore the space of possibilities, second to manage each personal creative approach. We learn from theoreticians in painting art about colours representation, organisation and uses. Mainly, we focus on Chevreul’s work about simultaneous contrast and the seven contrasts of Itten – hue, light–dark, warm–cool, complementary, simultaneous, saturation and quantity (Itten, 1977). The analysis of contrasts and colours in graphical works leads us to propose a definition and characterisation of colours harmony in cartography: we assume that a harmonious colours specification facilitates the reading and understanding of a map. Evaluations of linkage and balance between colours and of area/contrasts relations in a map have been proposed to measure the global colours harmony in a map (Christophe et al., 2011). This work permits to explore a part of visual perception of a map through the analysis of harmonious colours specification. Finally, any map-reader may select the relevant information in a map, if it is correctly built according to all main

141

Creative Colours Specification Based on Knowledge (COLLEG)

Figure 1. Map samples: colours uses respecting semantic relations

principles we have overviewed, depending on their own cognitive capabilities (eventual handicaps to manage) and contexts of use (eventual final supports and generalisation of symbol specification to anticipate). This overview of knowledge from various domains leads us to the necessity to formalize it with the help of basic rules. Proposition of formalized cartographic colours uses

From identified knowledge presented in the previous paragraphs, we extract some basic cartographic rules: 1. Semantic rules: ‘If Theme A and Theme B have a relation R, then their colours have a relation R’. 2. Contrasts rules: ‘If Theme A is cartographic background, then hue/value contrasts between A and the other themes have to be high enough’. 3. Conventional rules: ‘If the semantic of Theme is S, then Theme A should be rendered by a conventional colour for S’. These rules may be formalized as: ‘If (Concept 1) has a value xxxx then (Concept 2) should take the value yyyyy’. We use the definition of cartographic concepts we describe in Domingue`s et al. (2009). The premises of the rules mainly concern concepts related to the one of Theme: SemanticTheme related to the name of the theme: {‘sea’, ‘buildings’, etc.} and QualityTheme related to the role of the theme: {background}. Conclusions concern mainly the following concepts: Colour, Hue, Value, HueContrast, ValueContrast and ColourFamily. A ColourFamily is defined by a constitutive hue {blue, red, etc.} and associated hues {blue-purple, blue-green, etc.}: this concept supports the definition of conventional colours depending on the required context (Christophe and Ruas, 2009). We describe the six rules we use from (R1) to (R6), in considering two themes T1 and T2, and their related colours C1 and C2: (R1) (R2) (R3)

(R4)

If Relation(T1,T2)5difference then HueContrast(C1,C2)5high. If Relation(T1,T2)5association then HueContrast(C1,C2)5low. If Relation(T1,T2)5order then HueContrast(C1,C2)5low and ValueContrast(C1,C2)5 high. If T1.quality5background then C1.Value,3.

(R5) (R6)

If T1.name5sea then C1 in (ColourFamily5 Blue). If T1.name5wooded area then C1 in (ColourFamily5Green).

Proposition of inspiration sources and formalized artistic colours uses

In the context of assistance to make better colours choices, basic cartographic rules have to be faced to user’s preferences. We assume that building a colour specification from scratch may be difficult. We propose to stimulate users’ imagination with the help of inspiration sources and make them reasoning by analogy:

N N

map samples presented in Figure 1 describe colours uses respecting semantic relationships between geographical themes (association, difference and hierarchy) and thus adapted but not necessary conventional colours specification; famous paintings support artistic colours uses, with specific colours palette and painter’s rules (type of repartition and proximities of colours in the painting). Figure 2 presents a painting of Matisse, its colours palette and painter’s rules we extracted from.

These inspiration sources are knowledge supports on colours uses in visual representation. We would like to propose them to users in order to make them express their preferences on colours and on colours uses in a cartographic context. Our knowledge base is thus made up of inspiration sources, artistic and cartographic rules. Knowledge has now to be managed by an operational system.

KNOWLEDGE MANAGEMENT BY THE COOPERATIVE SYSTEM COLORLEGEND

Our proposition of a man–machine dialogue model helping a user to make personalized and original colours choices during map design process has been detailed in Christophe and Ruas (2009). This model has been implemented in the COLorLEGend (COLLEG) system. The system is responsible for five main tasks: constraints management, design, evaluation, decision making and repairing. Design task relies on a cooperative method in four stages, based on man–machine interactions: 1.

Choice of an inspiration source by the user.

142

The Cartographic Journal

Figure 2. A painting of Matisse, its colours palette and its painter’s rules

Figure 4. Representation of artistic constraints

N

csonventional rules characterized by the name of the theme {sea, wooded area, buildings, etc.} and the more suitable colours family {blue, green, yellow, red, etc.}.

Figure 3 presents the class diagram of the cartographic constraints. Artistic constraints

Figure 3. Representation of cartographic constraints

2. 3.

4.

User’s preferences acquisition from inspiration source. Interpretation of user’s preferences into constraints and making of possible colour specifications from current constraints (user’s, cartographic and artistic). Refinement of colours by the user.

Man–machine interactions allow the user to express various preferences and the system to propose various colours specifications. Moreover, they favour user’s creativity throughout the process. In this section, we specify that our knowledge base is represented and managed as constraints on the colours specification in progress. Constraints representation

The colour specification in progress is under various types of constraints: user’s ones, cartographic ones or artistic ones. To automatically manage colours, we rely on the colours reference system of Chesneau (2006) and Buard and Ruas (2009) and use their instantiations of colours. They are characterized by their hue, value and saturation. User’s preferences

The user may select a colour, for a specific theme or not: the system interprets it as a possible colour for the colours specifications in progress. Cartographic constraints

Basic cartographic rules presented in the previous section can be divided into:

N

semantic rules characterized by a type of relation between themes {association, difference, order}, and thus a level of hue and value contrasts {high, medium, low};

Paintings are represented by their colours palette and their painter’s rules as artistic constraints. Three main attributes are described: repartition type, mean surface and proximities. Figure 4 presents the class diagram of the artistic constraints. Constraints management by COLLEG

Constraints management is considered as user’s preferences interpretation, rules analysis and colours specifications making. The COLLEG system applies all possible colours to themes in the key and thus to objects in the map. It relies on a constraint satisfaction problem algorithm: a set of variable X1, X2, …, Xn (themes) may be rendered by a set of values x1, x2, …, xn (colours) that should satisfy a set of constraints C1, C2, …, Cn (cartographic and artistic). At the end, various maps are made highlighting all possible colours specifications, according more or less to user’s preferences, cartographic and artistic constraints. The user may then refine some colours, if necessary. They can use the refine tool of the system relying on current constraints: hue/value shades, conventional colours, colours palette are proposed to users to improve colours specification, according to their tastes. Cartographic and artistic variations for user’s data

From the painting of Matisse presented in Figure 2, we may obtain the two following maps in Figure 5, for a user’s dataset made of the following themes {sea, wooded area, background, buildings, main and secondary roads}. Colours palette of the painting is used. Respecting cartographic constraints involve that blue colours are applied to sea, green colours are applied to wooded area and lighter colours are applied to background. Other possible colours are applied to other themes, according to semantic constraints: relations of difference and association have to be respected (map on the right). Respecting artistic constraints involve that colours are applied according to painter’s rules, respecting the

143

Creative Colours Specification Based on Knowledge (COLLEG)

Figure 5. Two (most different) maps from Matisse

Figure 6. Examples of colours specification from a painting of Vincent Van Gogh (La Chambre a` Arles, 1888, Van Gogh Museum, Amsterdam, Netherlands)

repartition type and proximities of colours: this application requires that the system knows characteristics of area, size, shape and proximities of the cartographic objects to render (map on the left). From paintings, the user may obtain various maps according to their tastes and needs and to the respect of cartographic and artistic constraints. Figures 6 and 7 present four possible maps coming respectively from a painting of van Gogh and a painting of Derain. More or less original or realistic maps may be obtained in order to satisfy the user. COLLEG has been experimented by a user test: expert and novice users had to make a map according to their tastes, with the help of one of five paintings. Most of users found COLLEG easy to use, efficient to make an original

and cartographically correct colours specification and very helpful to select colours combinations. A result of this work is the possibility to manage colours through a 3D space defined by conventional, contrasted and artistic axes: various maps may thus be made according to user’s tastes and needs varying more or less on the first, the second and/or the third axes.

CONCLUSION, DISCUSSION AND PERSPECTIVES

In this paper, we handle with colours specification during map design process. Our purpose is to help users to make personalized and original colours specification while respecting cartographic theory. We highlight the complexity of this

144

The Cartographic Journal

Figure 7. Conventional, contrasted and artistic colours specifications from a painting of Andre´ Derain (Montagnes a` Collioure, 1905, National Gallery of Art, Washington, John Hay Whitney Collection)

problem and the requirement of knowledge coming from various domains such as visual perception, cognitive sciences, graphic semiotics, cartography and art. We specify cartographic rules (semantic, contrasts and conventional) and artistic rules that may be confronted. We propose to rely on inspiration sources (map samples and famous paintings) as knowledge supports to manage cartographic and artistic colours uses. Our choices of constraints (from system: cartographic and artistic, and from user: preferences) representation is managed by our COLLEG system. The COLLEG system allows users making original colours specifications according to their tastes and needs. At this step, needs are just evaluated by the user himself. Further work may improve the given assistance in proposing to users to respect specific constraints to render a specific geographical phenomenon. For instance, if the user needs to represent the extent of the roads network in their territory, COLLEG should warn him to respect the contrast rule while rendering roads: a high hue or value contrast has to be respected. Moreover further to the users’ test, users ask to see and to manipulate themselves the current constraints used or not for each colours specification. This option may help users to understand how maps can be so different and thus support various cartographic messages while using the same colours set. In the context of the appropriation of cartographic tools by various users and in the context of the improvement of experimentation around visual variables of Bertin, the identification and formalisation of a knowledge base on colours uses are important steps in order to improve the knowledge about cognitive processes underlying map design process. Further research on the role of colour in building and understanding of cartographic message is under progress.

BIOGRAPHICAL NOTES

Sidonie Christophe is Doctor in Geomatics and Engineer in Agronomy. Her PhD thesis handles with the proposition of a methodology to make personalized and original colours specifications. She made a postdoctoral degree in the Laboratory of Informatics of Grenoble about analysis of human behaviour and mobility patterns. She is currently researcher in the COGIT Laboratory and works on graphic semiotics.

NOTES 1

Google Maps, Ge´oportail, etc. Free software (QuantumGIS, gvSIG, etc.) and proprietary software (ArcGIS, MapInfo, GeoConcept, etc.). 3 Semiotics is ‘the science of studying the role of signs as part of social life’ (Ferdinand de Saussure, 1916, Cours de linguistique ge´ne´rale). 2

REFERENCES Bertin, J. (1967). Se´miologie graphique: les diagrammes, les re´seaux, les cartes, Re´e´ditions en 1973, puis en 1998, Editions de l’EHESS, Paris.

Creative Colours Specification Based on Knowledge (COLLEG) Brewer, C. A. (2003). ‘A transition in improving maps: the ColorBrewer example,’, Cartography and Geographic Information Science, 30, pp. 159–162. Buard, E. and Ruas, A. (2009). ‘Processes for Improving the Colours of Topographic Maps in the Context of Maps On-demand’, in 24th International Cartographic Conference, Santiago, Nov 14–19, CD-ROM. Chatterjee, A. (2006). ‘The neuropsychology of visual art: conferring capacity’, International Review of Neurobiology, 74, pp. 39–49. Chesneau, E. (2006). ‘Mode`le d’ame´lioration automatique des contrastes de couleurs en cartographie’, PhD thesis, Paris-Est University, Paris. Christophe, S., Zanin, C. and Roussaffa, H. (2011). ‘Colours Harmony in Cartography’, in 25th International Cartographic Conference, Paris, Jul 3–8, CD-ROM. Christophe, S. and Ruas, A. (2009). ‘A Process to Design Creative Legend On-demand’, in 24th International Cartographic Conference, Santiago, Nov 15–21, CD-ROM. Christophe, S. (2009). ‘Aide a` la conception de le´gendes personnalise´es et originales: proposition d’une me´thode coope´rative pour le choix des couleurs’, PhD Thesis, Paris-Est University, Paris. Domingues, C., Christophe, S. and Jolivet, L. (2009). ‘Connaissances ope´rationnelles pour la conception automatique de cartes’, in 20th Journe´es Francophones d’Inge´nierie des Connaissances (IC2009), Hammamet, Mar 25–29, pp. 253–264.

145 Fabrikant, S. I., Rebich-Hespanha, S. and Hegarty, M. (2010). ‘Cognitively inspired and perceptually salient graphic displays for efficient spatial inference making’, Annals of the Association of American Geographers, 100, pp. 17–29. Fabrikant, S. I. (2005). ‘Towards an Understanding of Geovisualization with Dynamic Displays: Issues and Prospects’, in American Association for Artificial Intelligence (AAAI) 2005 Spring Symposium Series: Reasoning with Mental and External Diagrams: Computational Modeling and Spatial Assistance, pp. 6–11, Stanford, CA, Mar 21–23. Itten, J. (1977). Art de la couleur, Editions Dessain et Tolra, Paris. Mollard-Desfour, A. (1998). Le dictionnaire des mots et expression de couleur – Le Bleu, CNRS Editions, Paris. Montello, D. R. (2002). ‘Cognitive map-design research in the twentieth century: theoretical and empirical approaches’, Cartography and Geographic Information Science, 29, pp. 283–304. Rosch, E. (1976). ‘Structural bases of typicality effects’, Journal of Experimental Psychology: Human Perception and Performance, 2, pp. 491–502. Simon, H. A. (1973). ‘The structure of ill structured problems’, Artificial Intelligence, 4, pp. 181–201. Treisman, A. and Gelade, G. (1980). ‘Feature-integration theory of attention’, Cognitive Psychology, 12, pp. 97–136.