* * Centre de recherche en géomatique

For a few years now, several development projects have been carried out in municipal contexts in order to make spatial information available on the Internet. An overall observation of the existing municipal Web sites obviously shows the great variety of the objectives at stake, and of the technological solutions implemented. Despite the increasing number of researches dealing with the democratization of e-information addressed to citizens and e-governments, it is still difficult to clearly identify the current privileged means of communication between cities and citizens on the basis of cartographic data. This difficulty is related to the absence of formal and effective frameworks to characterize and classify the various ways to diffuse geospatial information on municipal Web sites. On the basis of the above, the present research aims at remedying this ignorance by elaborating a new classification framework to effectively describe GIS-based Web sites in municipal contexts. The adopted strategy consists in analysing the already existing partial approaches of classification, in order to pursue with the development of a more comprehensive pragmatic classification framework. This framework is then the subject of an experimentation consisting of a detailed analysis of the contents and functioning of a hundred existing municipal Web sites in Canada. Finally, this experimentation makes it possible to draw initial conclusions regarding the usability of the new classification mode proposed, as well as to identify some further research pathways.


Résumé
Depuis quelques années maintenant, plusieurs projets de développement ont été menés en milieu municipal dans le but de rendre disponible sur Internet de l'information de nature géospatiale.Une observation rapide des multiples sites Web municipaux nous permet de constater la grande diversité des objectifs poursuivis et des solutions technologiques mises en oeuvre.Malgré les recherches de plus en plus nombreuses portant sur la démocratisation de l'information électronique vers le citoyen et le gouvernement en ligne, il demeure difficile de bien cerner les modes actuellement privilégiés pour la communication entre les villes et les citoyens sur la base de données cartographiques.Cette difficulté est liée à l'absence de cadres formels et efficaces pour caractériser et classifier les différents modes de diffusion de l'information géographique dans les sites Web municipaux.Suivant ce constat, la présente recherche vise à pallier à cette méconnaissance en élaborant un nouveau cadre de classification pour décrire efficacement les sites Web à base de SIG qui existent en milieu municipal.La démarche générale suivie consiste d'abord à analyser les approches partielles de classification déjà élaborées, pour ensuite concevoir un cadre de classification pragmatique plus complet.Ce cadre a pu faire l'objet d'une expérimentation concrète lors de l'analyse détaillée du contenu et du fonctionnement d'une centaine de sites Web municipaux existants au Canada.Finalement, cette expérimentation permet de tirer des conclusions initiales sur l'utilisabilité du nouveau mode de classification proposé, ainsi que de dégager des pistes de recherche ultérieures.

Understanding GIS-based Web Sites: A Municipal Stake
For some years now, many development projects and research works have dealt with the democratization of spatial information through its availability on the Internet (Plew, 1997).Besides, scientific literature reports the constant questioning on this topic of interest (Cartwright, 1999, Kingston, Carver, Evans & Turton, 2000, Niles & Hanson, 2003).This interest is also expressed through the expansion of a great diversity of solutions aiming at improving cartographic diffusion on the Internet (Andrienko, Andrienko, Voss & Carter, 1999, Doyle, Dodge & Smith, 1998, Huang, 2003, Huang, Jiang & Li, 2001, Kähkönen, Lehto, Kilpeläinen & Sarjakoski, 1999).
Regarding the more specific context of municipal organizations, the projects related to the diffusion of urban spatial information have also increased during the last few years (for instance: Gauna & Sozza, 1999, Wilmersdorf, 2003).Consequently, the democratization of this type of information is still at the origin of many stakes and questioning as for its probable avenues of future development, either within the organization (Hendriks 2000), or in the external relations with citizens (e-government) (Gandhi & Cross, 2001, Smith, 2001).A close look at the various online municipal sites obviously shows the current swarming context, for there is a significant quantity and diversity of such Web sites.In certain sites, the map is a mere static image, which only offers the localisation of the city in its region.In more complex Web sites, sophisticated dynamic maps can be found, constituting a privileged means to transfer municipal information to citizens.Moreover, certain Web sites appear as "juxtaposed" and subordinate compared to the city's traditional means of communication, whereas other Web sites make it possible to compromise with the municipal administration (e.g.: to fill in forms online) and thus appear deeply integrated into the city's business processes.It is still difficult not to get lost into such diversity of municipal Web sites, and of targeted purposes.
The question thus lies in the current difficulty to draw an overall picture of GIS-based Web sites within municipal contexts.More specifically, the actual diversity of GIS-based Web sites, regarding both their form (symbology, functions, etc.) and their content (e.g.themes) makes it difficult to determine their real utility for municipal organizations.This diversity is inherent to the recent convergence of two innovative communication modes between cities and citizens.First, in this e-democracy age, the increasing use of the Web as a way to communicate in a municipal context both completes and challenges the traditional communication modes and services delivered to citizens.Second, even if municipalities' managers and citizens have used maps for quite a while, maps are now in a digital form and can come with sophisticated display and analysis functions.Nowadays, the convergence of these two communication modes, the Web and digital maps, allows an easy diffusion of online dynamic digital maps.
This emerging context creates a situation where the number of GIS-based Web sites is growing rapidly.Thus, we need to understand the relevance and the contribution of these new communication modes.To support this comprehension, it is essential to effectively describe these types of Web sites first.But to do so, there is actually a difficulty related to the absence of formal and effective framework to characterize and classify the various means to diffuse geospatial information onto municipal sites.In the light of the above, the present research aims at remedying to this ignorance by investigating this aspect, in order to elaborate an innovating and effective classification framework to describe GIS-based Web sites in municipal contexts.In a pragmatic perspective, we think this classification will eventually allow municipalities' managers to identify the level of e-democracy of their city related to others (caparison approach) and to find the best solutions to implement in their context (contingent approach).
In order to construct such a classification framework, the adopted strategy consists in analyzing the current classification approaches that can be found in the existing scientific and professional literature on the one hand, and in designing a classification framework on the other.Finally, the new framework is concretely tested by analysing the contents and functioning of a hundred existing municipal Web sites.
In what follows, we first present the theoretical background and the research methodology on which the development of the new classification framework is based.Next, we design the new framework, as well as the inherent classification criteria.Then, we experiment this new framework on the municipal Web sites of the province of Quebec (Canada), and valid it in the French context.Finally, we draw the appropriate conclusions, and propose further research pathways.

Theoretical Background
This section aims at presenting a short review of the literature relevant to our study.Thus, we first draw up an overall picture of the main thematics covered by literature.Next, we synthesize the existing technological possibilities to exploit online geospatial information.Then, we sum up the various cartographic formats present on the Web, as well as the various groups using this spatial information.Finally, these elements of information enable us to identify the existing types of classification, and to see how to complete them, in order to come to a more comprehensive structural framework.

Overall view of the literature dealing with municipal Web sites
A close look at the existing professional and scientific literature shows that there are only very few works and articles dealing with the diffusion of geospatial information on the Internet.Among these works and articles, those focusing on the online geospatial information within municipalities, are even rarer.The treated problematics are, most of the time, limited to the information technology aspect, and ignorant of geospatial informational considerations, and strategic aspects related to the implementation of such sites.Moreover, the works dealing with online geospatial information especially concentrate on the technical problems (e.g.Online GIS) (Gauna 1999, for example), and not enough on the objectives, relevance and form of the geospatial representations that municipal Web sites privilege (Bachy et al. 2003).
In addition, the growing usefulness of the Web to manage the territorial administrations stimulates some researches to better understand the value of the Internet for cities, and its power to improve relationships with citizens.This problematic has become a topical issue for some years now, (Attilio 1998, Mandelkern 1999, Carcenac 2000).On the other hand, the contributions in this field do not linger long on the geospatial aspect of the online processing of municipal data, and rather focus on the digital public participation, as well as on its organisational impact for local governments.
Several studies highlight the upheavals (particularly the advantages) brought by the new participative approaches (Kingston et al., 2000): public meetings with less constraints regarding specific place and time, access to information from any computer equipped with an Internet connection and this, 7 days a week, and 24 hours a day ("24/7 ", according to Kingston et al., 2000), possibility for the citizen to give his opinion anonymously, etc.On the other hand, as Blomac (2001) underlines it, the idea is not to break contact with the citizen, leaving him alone in front of his computer to get the answers to his usual questions.The Internet should propose new services that will facilitate information inquiries for citizens, for example, to better understand a land planning or land use project, which offer them grater possibilities to participate by asking more enlightened questions at complementary public meetings (VanderMeulen, 2002).

The various possibilities to exploit online geospatial information
MacEachren's framework (1995) is interesting (Figure 1) even if it does not correspond to a classification approach explicitly dedicated to the "online" aspect of geospatial information.Indeed, apart from the Public-Private axis of this three-dimensional frame model, it appears that a communication objective generally implies a weak Human-Map interaction.On the contrary a spatial analysis objective will imply a stronger interaction, and thus, the use of more sophisticated functionalities.It is relevant to establish such connections for the present study, and they will be taken into account while elaborating the new framework.According to Plew (1997), there are four different ways to exploit online geospatial information, but they are not all appropriate for the general public: -Simple presentations: certain sites may offer some geospatial information in text or chart format; -Digital maps: static, interactive or dynamic, they constitute a very efficient means to transfer geospatial information; -Raw data: for example, ESRI's "coverages" can only be used and handled by GIS (Geographic Information Systems) professionals; -Web-based GIS: online functions and data enabling queries or analyses (spatial or statistical), the creation of customized maps, etc.
Among these four types of geospatial information, two are exploitable by the general public: simple presentations and digital maps.Thus, the democratization of the use of raw data and Web-based GIS still seems limited.Already in 1993, Innes & Simpson (1993) affirmed that GIS were above all intended to experts, then accessible to professionals, and finally to the general public.Yet, there is a progression in the use and usefulness of GIS by the general public, even though it still remains marginal today.Indeed, it appears that the consultation of Web-based GIS is mainly occasional, and essentially involves experienced users, as government agencies, universities and industrial groups (Peng & Yu 2002).The studies showed that only 3.9% of the persons surveyed use Web-based GIS outside their office work, i.e. for their personal needs (Peng & Yu 2002).These results stress one more time the fact that this new technology remains confined to the community of GIS professionals, and has not yet penetrated the general public's skills.As for raw data, they are especially used by professionals wishing to supplement their own databases.

Online maps formats
Like traditional GIS, cartographic data can be transferred on the Web in two (2) types of basic formats: Raster maps and Vector maps (Longley et al. 2001).
A raster map is, in fact, a static image, a "photograph" of the map.It is an image represented by a grid of pixels that, once associated to one another, form the visual document.According to Finnseth and Jökulsson (2001), a map constituted of an image in raster format (TIF, GIF, JPG, PNG, etc.) is not considered as a dynamic map, even if it has zoom and pan functions.Raster maps generally have a fixed resolution, which may entail a bad visual quality when zooming in too much.Raster maps proposed on the Web can be generated in three different ways: either by making a high resolution scan of a paper map or an aerial photograph; or by integrating a satellite image; or by generating a raster map using a GIS or a CAD (Computer-Assisted Drawing) software.The raster format result can be then transferred on the Web site, just like any other type of digital image.
A vector map consists of the arrangement of graphic objects as the lines, points and polygons (Longley et al. 2001).Each object is autonomous and is identified by characteristics relating to its size, form and colour.The advantage of this format lies in the fact that the resulting maps are resizable without degradation.As Alameh (2002) underlines it, the Internet has brought GIS to the fore as distribution platform of online cartography.Yet, the problem is that there is still no standard vector format on the Web; thus each publisher of GIS software proposes his own diffusion vector formats on the Web.Nevertheless, two vector formats currently dominate imagery over the Web: the Flash format (most widespread) and SVG format (Scalable Vector Graphic).The Flash format is proposed by Macromedia, which requires a free plug-in that has to be installed in the Web browser of the client computer.Much newer on the market, the SVG format offers all the functions contained in Flash, but remains an open format.Specified by the World Wide Web Consortium (W3C), a body regulating the Web for the establishment of new standards, this format also implies the installation of a plug-in.

Types of Web users interested by online geospatial information
Keeping in mind the four online geospatial information presentations (simple presentation, raw data, maps and GIS), we may assess that different types of users are not interested in the same types of information.Therefore, three main user groups can be identified as follow (Plew, 1997): -GIS professionals: especially interested in the available geospatial data sets, they want to enlarge their own databases.These users (e.g.geomarketing specialists, researchers, engineers) are not interested in finalized maps, for they wish to use the available online data in order to generate their own spatial representations; -Public and private organizations: this type of user of geospatial information released by municipalities gathers the public administrations and other municipalities, contractors, governmental agencies, as well as citizens.These users are interested in any kind of geospatial information available, and relative to different thematics: economic, social and demographic data.Public organizations are often more interested in raw data, as in Web-based GIS, whereas citizens are more attracted by maps, or even Web-based GIS in the case of public participation (Kingston et al., 2000); -The general public: finally, every Web user can be interested in geospatial information offered online by municipalities.These occasional Web users may be potential tourists interested in certain regions and their local assets, or people wishing to move in the city.This mass public is more attracted by simple geospatial information (e.g.basic maps, pictures showing wide urban zones, diagrams locating a city in its region).Considering these different groups, it is obvious that the utility of online digital maps as means of communication will be different.The average user will want simplified geospatial information that will allow him to localize the city within the region or to localize a definite place (e.g.address, street) within the municipality.Moreover, this group of users might want to aim towards tourist information (e.g.localization of museums and monuments) or vacation information (e.g.parks, restaurants, bicycle paths network).Others, like territory managers (e.g.land surveyors, civil engineers, real estate appraisers, brokers and agents) will seek more precise information (e.g.residential urban zone of low density) or specifics topics (e.g.water network) related to their work.In addition, private organizations might be looking for information on infrastructures or on programs that apply to a city's industrial parks.. Finally, regarding the use of urban digital maps, the involvement of citizens in urban development projects results in the development of Public Participation GIS (PPGIS), a kind of interactive GIS-based Web sites, where opinions on territory management can be directly posted and debated over a Web interface based on digital maps (Roche, Sureau & Caron, 2003).

Existing classification modes
In the light of our bibliographical review, we have come to the fact that there are only few efficient classifications of Web sites.More specifically, a global classification of the municipal Web sites based on geospatial data is purely not to be found.Our approach is thus to develop a comprehensive classification of such sites, basing ourselves on several authors' works.
Therefore, we have thoroughly analysed different partial classifications, and others that had not been elaborated in the context of municipal Web sites.For instance, we have gone through Kosiur's (1997) and Peng and Tsou's works (2003), respectively for Web site classification, and for the classification of digital maps interactivity on the Internet.We also have enriched the concepts studied thanks to the results issued by the PhD of the fourth author of this present paper.In section 4, we use all these works to develop the new classification proposed.

Research Methodology
Initially, we especially wanted to base the design of the new classification framework on bibliographical review.But we soon understood that the current existing bibliography is still too limited to do so.Consequently, we have decided to complete and strengthen our theoretical approach with a concrete experimentation, that is to say, the analysis of existing municipal Web sites.The idea is to add concrete and precise measure criteria to operationalize the concepts that can be found in literature and to validate the coherence and easy use of the framework proposed.
For this reason the research approach followed is both very pragmatic and iterative, based on a literature review and on the analysis of real cases found in municipalities.In concrete terms, we followed these steps: • A review of the existing literature clearly identified the current Web sites classification modes, and sketched out a preliminary global classification.• In parallel, the initial analysis of several existing Web sites enabled us to generate a typology.• This cross-approach led us to produce a second version of classification based on accurate classification criteria.• This version has been readapted and improved after the study of twenty more sites.• Finally, the relevancy and efficiency of the new framework has been more thoroughly confronted to genuine "field data", that is to say, to the analysis of the content and functioning of a hundred of real municipal Web sites.This experimentation has made it possible to draw conclusions relating to the capacity of the framework to classify efficiently all the cases encountered, as well as to identify ulterior research pathways.Since the resulting classification is based on criteria "grounded" on many actual Web sites, we think the overall result represent a useful tool to help urban managers to identify the sophistication level of their municipalities' Web sites.
Regarding population, the analysis targeted all the city municipalities' Web sites of the Province of Quebec (Canada).As towns are not the same size in this Canadian province, we have decided to focus on middle-size to big cities, that is to say "city municipalities".We assumed that a larger percentage of these municipalities may propose Web sites, compared to other types of municipalities (e.g.villages).After an initial examination, we have found that among the 198 city municipalities of this province, 97 proposed a Web site.Our research, thus, focused on these 97 sites, which we analysed in April and May 2003.

Development of the Classification Framework
More generally, municipal Web sites based on geospatial data are… Web sites!Hence, the classification axis that naturally comes to the fore is based on the goal of these sites.The overall objective, which fundamentally guides the creation of a Web site, directly impacts on the richness of the site, in terms of functions.
Moreover, the content of the sites relating to municipal organizations is inevitably strongly focused on the territory managed by these municipalities.That is to say, the territory and its cartographic representations are likely to constitute a fundamental aspect of such Web sites.Consequently, the classification of these cartographic representations stands out as a second essential classification axis.
In what follows, we develop these two axes on the basis of existing works.We also carefully establish detailed criteria based on the analysis of Web sites, in order to eventually operationalize each axis.Finally, we assemble both axes to develop the proposed framework.Kosiur (1997) proposes different types of commercial Web sites, by classifying them into four (4) categories, which are, in their turn, defined by three (3) characteristics:

Existing Web sites classifications
-Publishing: advertisement, marketing, information.
-Processes: executions, regulations, tasks and operations.Kosiur (1997) underlines that each category constitutes a maturation step on a linear scale.Thus, an organization starts with a simple Web site to progressively move towards a more sophisticated type of site, as well as more integrated to the organization's functioning.This maturation of Web sites is thus related to the richness of functions offered online (Figure 2).Basic functions (Publishing) consist in publishing information related to the services and products that are being available online.Interactivity enables Web users to register with an e-mail address.The "Transaction" category makes it possible for the site user to interact with a database to proceed to online payments.Lastly, the ultimate phase (Processes) consists in having a Web site which content and functions are fully integrated into the organization's business processes.

Figure 2: Evolution of business functions on the Internet
Source: Kosiur 1997, p. 211 In the light of the preliminary analysis of the different municipal Web sites (about twenty), it is clear that all the sites easily fit one of the four maturity levels proposed by Kosiur (1997).In fact, there is an important variability in the functions offered to the citizens, from one Web site to another.Taking this fact into account, we consider that Kosiur's classification is generally transposable to the municipal context.Moreover, we believe that the "maturity" logic presented by Kosiur (1997) is also applicable to municipal Web sites.Indeed, the simplest sites propose only information, and advertise their region.Other sites offer to register with an e-mail, or with online forms.The most elaborated sites propose to citizens to register directly in the municipal databases by completing teleprocedures.Finally, even if it is difficult to make an accurate assessment without any local case studies, certain Web sites seem to be closely integrated to the municipal administrative processes (exchanges with other territorial organizations, products or services available on the Internet, etc.).

Proposed classification axis of the Web sites
According to the facts introduced above, and on the basis of Kosiur's (1997) criteria, we have analysed if the Quebecer municipal Web sites had predominant characteristics, in order to classify them into one of the four types of Web sites.The gathering of these characteristics should lead us to the identification of the criteria, which in turn will enable us to operate an efficient classification of the municipal Web sites.
The classification has been elaborated on an encased pattern (like Russian headstocks: Matryoshka), that is to say growing according to the maturation of the available functions.Therefore, in order to define the classification criteria, we had to assess from which level the presence of a certain characteristic was a discriminating criteria.For instance, a transactional site is able to offer at least the same functions as an interactive site.Consequently, the functions of the inferior sites are not excluded as we progress within the classification, but they are simply integrated.
After several iterations, five (5) significant criteria have finally been adopted to discriminate the types of Web sites, which are as follow: Criterion 1. Site with promotional information or offering administrative documents Criterion 2. Unidirectional or bidirectional site Criterion 3. Site with or without online procedures Criterion 4. Site with or without online pay products or services Criterion 5. Site requiring a reorganization of the municipal administrative management Criterion 1: site with promotional information or offering administrative documents.A site which only purpose is to advertise the territory is automatically classified as a purely informational site.Yet, as soon as the administrative information enables the local governments to interact with their citizens, the site is at least classified with the interactive sites (if not with the transactional or integrated sites).
Criterion 2: unidirectional or bidirectional site.Another criterion to characterize the "informational" type of the other kinds of sites is to check if the site has for only purpose to be read (unidirectional), or if the Web user has some possibilities to react regarding the content of the Web site (bidirectional).In this latter case, we would then try to find out if the Web user has the possibility to send back some information to the municipality, for instance via a specific e-mail address given by the municipal services, or via some printable forms or teleprocedures (in the case of transactional sites in this latter example).
Criterion 3: site with or without online procedures.The presence of teleprocedures is absolutely characteristic of the transactional sites.Teleprocedures are digital forms, which can be filled in and transferred online.If the form has to be downloaded from the Web site, printed, and then sent via the standard postal services, the site is said to be only interactive.
Criterion 4: site with or without online pay products or services.As soon as products or services can be paid directly on the municipal Web site, the site is at least transactional (if not integrated).Informational and interactive sites are not expected to propose this type of functions.
Criterion 5: site requiring a reorganization of the municipal administrative management.This kind of Web site is more difficult to identify.It requires knowing if the functions offered by the municipal Web site are not merely "juxtaposed" to the traditional municipal services, or if a reorganization of the administrative processes has really enabled the integration of the e-government mode to the existing habits.As the limit between a transactional site and an integrated site is relatively thin, it may be necessary to have a closer look at the functioning of the municipality directly on the spot.
Combining all the five bimodal characteristics, it becomes possible to define a classification axis based on the discriminating criteria observed (Figure 3).Peng & Tsou (2003) propose a classification of online digital maps, according to an ascending scale of the level of interactivity and of the richness of the available functions:

Existing classification of online digital maps
-Static Map Publishing: static image map -Static Web Mapping: maps displayed according to certain predefined parameters -Interact Web Mapping: maps offering richer functions regarding the customer application: zoom, pan, queries, choice of the layers to be displayed -Distributed GIS Services: online access to most functions as: query, analysis, manipulation and display of a standard GIS.
This gradation progressively goes from a mere static digital image to a complete package of functions that are generally to be found in GIS (Figure 4).Moreover, Capt (2002), who had adopted a categorization from C. Harder (Serving maps on the Internet: geographic information on the World Wide Web), worked on the elaboration of a typology of online maps, constituted of five (5) categories: -Localisation maps -Maps which content changes -Interactive maps -Maps resulting from a spatial analysis -Maps undergoing a geospatial process Within this classification, localisation maps are described as being "the digital equivalent of the paper plans.[…] These plans are often scanned from paper plans, or worked over again from vectorial files."This definition tends to make synonymous "static maps" with "localisation maps".Yet, every localisation map is not static, and every static map is not a localisation map.Another limitation to this classification is linked to the fact that it doesn't take into account hyperlinks possibilities, excluding thus certain types of maps.Finally, this classification doesn't bear in mind the possibility to offer advanced functions of Web-based GIS.
As far as they are concerned, Finnesth and Jökulsson (2001) make a distinction between static and dynamic maps.Even though these authors term "dynamic" all the different kinds of interactive maps, they underline the fact that having the possibility to perform zoom or pan operations on the map, is not a form of dynamism if the cartographic content does not change.According to them, the main characteristic of a dynamic map is that the content may change.Although this distinction helps us to distinguish dynamic maps from static maps, the authors do not give any detail about how identifying the various degrees of dynamism inherent to interactive maps.
In their book, Cartwright, Peterson & Gartner (1999) took interest in one specific type of dynamic maps: hypermaps.A hypermap is a particular form of hypermedia document.This type of map allows the Web user to use hyperlinks to have access to complementary information (texts, pictures, graphics, etc.) from a map, creating thus a particular form of interactivity.Consequently, this hypermap concept, as defined by its authors, constitutes a specific type of interactive map.On top of this online map category, the authors emphasize other forms of interactive cartography, and term all of them "multimedia cartography".
Kennedy and Barclay (1996) add complementary definitions to those proposed by Cartwright et al. (1999), which bring some more light to the multimedia concept.They make it possible to identify a sub-type of multimedia-dynamic maps, corresponding to the concept of "animated map" proposed by Caquart (2001).
Apart from the types of map introduced above, we have observed that certain Web sites offered an even greater level of interactivity, providing numerous functions that can also be found in GIS.In this case, Web users are not only people consulting online maps, as they also can directly act on the informational content of the maps.This kind of dynamic maps is generally based on "engines" providing the functions of Web-based GIS (e.g.Arc-IMS from ESRI or JMAP from KHEOPS).

Classification axis of online digital maps
Leveraging on the different classifications analysed, maps are mainly classified according to their level of interactivity.We can thus consider a gradation going from simple and static maps, up to a very dynamic level of consultation, where various functions of cartographic display manipulation are available to the Web user.In the light of the above, we propose a classification of the cartographic representations based on four (4) levels, progressively going from a static mode towards a dynamic mode: -Static maps -Hypermaps -Dynamic maps -Web-based GIS As for the Web site classification, the one proposed here has been elaborated on an encased approach.Therefore, the number of functions available to the Web user increases as the degree of interactivity progresses on digital maps (Figure 5), which is in keeping with MacEachren's idea (1995).

Figure 5: Proposed classification of online maps
In order to define the classification criteria, we had to assess from which level the presence of a certain characteristic was a discriminating criteria.For instance, a dynamic map offers at least the same functions as a hypermap.
After several iterations, seven (7) significant criteria were finally adopted to identify the digital map types on the Web.A map is said to be "dynamic" if its visual display can be modified by the Web user.Consequently, four (4) criteria enable to assess this modification of the visual display, and thus to define if the map is dynamic or not.The seven (7) criteria are as follow: Criterion Criterion 2: maps with descriptive data, or none.The purpose is to determine if the descriptive information is linked to the proposed map (e.g.hyperlinks, interrogative query).Such complementary information may also be accessible by hypertexts, clicking on geospatial objects located on a map, or by queries (on dynamic maps).Static maps do not propose such complementary information directly accessible on the map.
Criterion 3: customizable map, or not.A dynamic map, or a Web-based GIS, allows Web users to create customized maps.The user may choose the information layers he wants to be visible (to select layers in a preset list in caption, to change the display order, etc.).The symbology may also be modified by the user (colours, forms, canvas, etc.).A hypermap does not offer this display flexibility.
Criterion 4: maps with or without multimedia animation.This category makes the difference between dynamic maps and hypermaps.The maps offering visual or audio animations are automatically classified into the dynamic map category.
Criterion 5: maps allowing queries.Hypermaps send to descriptive information (text) via hyperlinks only.If it is possible to get information via queries, then the maps is dynamic, or it is a Web-based GIS.The queries used to search addresses or places, and then to see dynamically the result on the map, are the most common queries.
Criterion 6: maps with successive zooms delivering more and more details.In the case of dynamic maps, the successive zooms enrich the maps with additional layers of more comprehensive information.On a dynamic map, the successive zooms generate maps that are always accessible on the same portion of the Web site (same zone on the page).By contrast, a hypermap may allow zooms on certain portions of the maps, but the new maps are then accessible on other portions of the Web site.
Criterion 7: maps generated from manipulable geospatial data, or not.In addition to the visualisation functions of dynamic maps, Web-based GIS allow to add, modify or delete geospatial data.It gives access to the descriptive data on which the digital map is structured, to modify the value of these data, and consequently, to modify the content and form of the resulting maps.This "writing" access of the database enabling the creation of the map is the characteristic which makes the difference between a Web-based GIS and a dynamic map.
Combining all the seven bimodal characteristics, we can develop a classification axis based on discriminating criteria (Figure 6).

Classification framework combining both axes
Now we have elaborated criteria enabling to determinate the degrees of maturity of these Web sites, and to identify the types of maps distributed by local governments, we may develop a global classification framework.It will constitute an operational tool to classify municipal Web sites, that is to say, a classification matrix resulting from the combination of the types of Web sites (x-axis) and of the types of online digital maps (y-axis).A Web site is thus to be identified among 16 possible categories, going from A to D regarding the level of sophistication of Web sites, and from 1 to 4 regarding the level of interactivity of digital maps (Figure 7).Therefore, a Web site classified as A1 would point to a municipality hardly involved in the diffusion of geospatial information on the Internet, whereas a site classified as D4 would point to a municipality being on the leading edge of the current means of diffusion and interaction on the Internet.This matrix is interesting as it is capable of synthesizing a specific situation consisting of many GIS-based Web sites.Actually, despite its 16 possible categories that may appear complicated at first glance, a global analysis of the matrix offers a broad view of the situation (Figure 8).Moreover, such a matrix could be used to determine if there is any longitudinal pattern in Web sites progression.Consequently, and even though the following comment strays a little off the subject of our study, we do believe that the matrix could efficiently reveal if Web sites evolve, with the passing of the years, from the lower left corner towards the upper right corner of the matrix, that is to say, towards more and more sophisticated GISbased Web sites (Rogers 1995).

Experimentation
The objective of this experimentation is twofold: first of all, to determinate the internal qualitative validity of the classification matrix proposed, and secondly, to assess its external qualitative validity.Keeping with the first objective, we wish to analyse existing municipal Web sites, in order to check the efficiency of the various classification criteria to identify accurately each of the 16 types of the matrix.Keeping with the second objective, we wish to assess the global efficiency of the matrix to draw an overall significant picture of a specific organizational situation, which in our study corresponds to "city municipalities" of the Province of Quebec, (Ministère des Affaires Municipales et de la Métropole 2002).With respect to our study, this external qualitative validity will only be the object of a first estimation on the part of the researchers, and will not be measured compared to other types of organizations; further researches will be required to validate this estimation.

Analysis approach
To carry out the analysis, we have first created an evaluation grid based on the categorisation criteria inherent to the classification matrix.This analysis grid, consisting of questions/criteria, provides Boolean answers (True or False) or multiple-choice answers.Table 1 shows a sample of the detailed grid used to classify the numerous studied Web sites.In order to lighten the text, only a few criteria are being presented in this sample, and they only relates to the ten bigger municipalities studied.

Table 1: Sample of the analysis grid used to classify GIS-based Web Sites
In order to analyse the municipal Web sites with this grid, we have chosen the following approach.For each Web site, we first inquire about the questions/criteria linked to the types of Web sites, and to the simplest types of digital maps (type A1).If there is at least one positive answer to one criterion at that level, then we do consider that this level has been "reached", and we proceed to the "superior" one, that is to say, to inquire about the questions/criteria of the types of Web sites that are more on the right side of the matrix (towards D), or of the types of digital maps that are higher on the matrix (towards 4) of Figure 6.If all the answers are negative to reach a specific level, first of all, we do not try to proceed towards a superior level, and secondly, the Web site is identified to belong to the inferior level.
For example, a site answering negatively to all the criteria of the category "interactive site" (Interactivity) is automatically registered in the inferior category, that is to say an "informational site" (Publishing).But if, at least, one answer is positive, we may consider that the site is at least an interactive site, and if all the answers are positive at that level, the criteria of the following level are being tested, that is the "transactional site" level.

Global Statistics
Among the 97 cities' Web sites of the of the Province of Quebec, 8 point to wrong URLs (the sites were not studied in this case), 4 sites are under construction and 1 points to the County Regional Municipality's Web site from which it belongs (territorial grouping on a regional level).The analysis of the cities will thus only be carried out on 84 Web sites (Table 2).
Table 2: Distribution of the studied Web sites according to various initial facts Among the 84 Web sites, one proposes a digital map through another site dedicated to generating maps (MapQuest), and 7 do not offer any digital map.Actually, at least one online map is available out of 76 Web sites (Table 3).

Table 3: Amount of the Web sites containing online cartography
These 76 Web sites offering online maps correspond to cities with an overall population of 4.9 million.This reveals that theoretically, 67% of the population of the Province of Quebec have access to geospatial information relating to their municipal territory through the Web.

Analysis of the types of Web sites in a municipal context
In the light of the analysis of these 76 Web sites, we have come to comprehensively better understand the characteristics of the four levels of Web sites that are most likely to be found in a municipal context.
Informational sites offer only little services.They distribute only general information relating to the municipality for non-targeted users.The information may be consulted by any Web user who is interested in this municipality: tourist, citizen, businessman, property developer, etc.Such sites deliver information on various subjects: history, geography, region, economy or demography.Moreover, an overall presentation of the municipal services and departments can be found there.These Web sites also often tend to advertise their municipality (tourist, cultural, environmental attractions, or others related to the quality of life).When the sites are tourist oriented, they only aim at highlighting their city.No administrative information is available; the main objective of the site is not to offer e-services to citizens, but only to advertise the municipal territory.Informational sites barely exploit the interactive possibilities of the Internet.Eventhough most of the Web sites propose a generic and single e-mail address to receive questions and comments regarding the municipality, this basic function is not considered as being a real interactive service.Moreover, we have noticed that this type of generic e-mails was often badly exploited by the municipalities, who can take several weeks to answer, or even never answer at all.
Interactive sites enable municipalities to give access to information and documents regarding the municipal life or administration.Citizens and contractors are the main type of Web users targeted.The information available give details about community life (waste removal schedules, access to the different activities…).They also may concern municipal regulations, printable forms/questionnaires to be sent back by post (or directly at the counter of the town hall).Additionally, this type of Web sites may enable citizens to send back information to the various municipal departments through targeted e-mail addresses (urban planning services, public works, clerk of the court, etc.).All in all, these sites not only propose information, but also basic online services preventing the citizens from moving up to the town hall.
Transactional sites allow citizens to intervene directly in the municipal databases.In this case, e-forms are not to be printed, but filled in and sent online; they are teleprocedures.A teleprocedure is an e-document that transits on a network, where it has to be updated (completed).Finally, the e-document reaches its destination in the same digital format, to be used as such (Risson, 2002).E-documents, thus sent back, enrich directly the database without any intermediary manipulation from the staff.Municipalities' transactional sites also allow citizens to pay directly online for services or products.This type of service perfectly fits the collection of municipal taxes, electricity bill, etc.We have observed that municipalities could also sell products dedicated to their own promotion (clothes, pens, key rings with the town's coat of arms).
Integrated Web sites are deeply rooted in the municipality's administrative life.Their creation may have required a certain type of reorganisation of the municipal departments.The online services enable the municipal staff to work differently.With this e-government approach of the municipal management, citizens are oriented in priority towards the Web site, rather than to the town hall.The services are thus not only a complement to those supplied at the town hall, but are an integral part of the administrative processes.In this case, many products and fully digital services are deliverable from the Web site.According to our experimentation, this type of sites remains quite difficult to identify.On the Web site, the municipality may suggest that for better service, citizens should use the services available online in priority, rather than coming up to the municipality's counter.

Analysis of the various types of digital maps
On the same analysis basis, that is 76 Web sites selected, we have come to better apprehend how the four types of digital maps found on municipal sites, could be concretely characterized.
Static maps in municipal Web sites are often used to locate a city in its region.These sites often deliver maps in PDF format of raster or vector type.In a PDF document, the zoom and pan function are always available.With a raster map embedded in a PDF, too much zooming entails a pixelized picture.On the contrary, a PDF vector map prevents this trouble.PDF vector maps allow word searches.Yet, this option is not characteristic of an interactive query since it does not use the data base sub-jacent to the map, but only an optical character recognition.
Hypermaps make it possible, from a global map representing a municipality (e.g.urban zones), to click on a specific zone to get a new detailed map (larger scale) predefined of this zone, generally on another Web page.In this case, the global map presenting the territory is often divided into several hot zones that are invisible to the user, and constituting hyperlinks towards further information relating to these zones (texts, pictures, videos…).Most of the time, these hot zones are attached to a "mouse over" function, in other words, when the mouse pointer moves on one of the clickable areas of the map, a temporary label displays a brief comment.
Dynamic maps are maps which content and visual display can change whenever the user takes actions (Finnseth & Jökulsson 2001).Kennedy & Barclay (1996) explain that a dynamic map can be identifiable when a user can proceed to several successive zooms without being sent to other maps on other Web pages (which is the case with a hypermap); the Web page (URL) remains the same.Moreover, municipal maps resulting from successive zooms are not "predefined" as hypermaps are.When the user proceeds to successive "zoom in", new information appears little by little, usually implying queries in the database between the end-user's computer and the server.For instance, a global map of the metropolitan region may only show the principal roads and the urban areas, but zooming in progressively reveal the boulevards, names of parks and squares, subway stations, etc.The dynamic map also enables the user to customize his map by activating/deactivating within the legend the appropriate cartographic layers.He can also modify the visual attributes (colour, form, texture) of the geospatial elements.Finally, we have observed that certain municipal dynamic maps propose visual and audio animations (often simply by moving the mouse pointer above a cartographic element).
Web-based GIS enable not only to see geospatial information on the Web, but they also allow customized acquisition, manipulation, management and display of geospatial data.We shall precise that we have not identified any of the municipalities using this sophisticated type of technological infrastructure on the Internet.Yet, it does not mean that this level of dynamism is not being used within municipalities, for we do know that this type of online digital cartography is used in intranet mode by some large-scale towns, but the online access is denied for the general public.

Overall picture of the municipal context studied
Half of the Web sites (37 out of 76) propose static maps, which correspond to level 1 (Figure 9).Only 13 cities propose online dynamic maps (level 3).As far as we have observed, no Web site propose any online Web-based GIS.Actually, we have noticed a decreasing number of maps as we moved towards the four levels of dynamism of maps: 49% of level 1 (Static Map), 34% of level 2 (Hypermap), 17% of level 3 (Dynamic Map) and 0% of level 4 (Web-based GIS).It seems that there is a direct correlation between simplicity, and online cartographic diffusion technology, and the number of cities adopting this technology.We have also observed that a large proportion of the cities' Web sites is at least of the interactive type (93% of type B, C or D).Only 7% of the sites are just informational (A -Publishing).The dominant type of Web sites is type B (Interactivity) representing 62% of all the analysed sites.These results reveal that most municipalities have got beyond the initial stage, which is the use of the Web as a mere static "Business Card", and exploit multimedia technologies to interact with citizens through basic interaction functions: access to information regarding municipal administration, service delivery, recreational activities, printable forms to be sent back by post, specific e-mail addresses, etc. Generally speaking, the most part of GIS-based Web sites in municipal contexts (55%) matches type B1 or B2, which corresponds to interactive with static maps or hypermaps.

Validation with another context: France
So to carry out an additional validation relating to the efficiency of the new classification matrix, we experimented with its use in a second context.We chose to analyse the French municipalities in a very similar manner to that described above, for the study of Canada (Quebec).Therefore, we will not describe it here.The analysis was carried out between April and July 2003, and the sample was established from the following address: http://www.citaenet.com,where the 3353 French municipalities' official Web sites are listed.Solely the prefectures of the metropolitan departments (total of 95) were selected, which account for most French cities.Once the sample was submitted to the iterative approach presented above, 74 Web sites, that is to say a little more than 77%, were selected for the systematic analysis.22 million of citizens are concerned by these 74 Web sites, which represents 39% of the French population, homogeneously spread out on the national territory.
The matrix below (Figure 10) highlights the fact that 40% of the Web sites propose static maps (31 sites out of 74), while only 22 out of 74 propose online dynamic maps.Solely two Web sites offer "online-GIS" type functions: Strasbourg and Laval.They both use a fairly new technology hardly exploited by municipalities, even though some weaknesses remain, like the impossibility to modify the content of the maps online.Yet, taken as a whole, French cities' online maps offer a higher level of interactivity than those released by Quebec towns.Most of the Web sites developed by French cities rank among interactive sites, of type B (36 out of 74), which is the second level of evolution.It should also be noted that 30 informational sites rank at the first level of evolution, which means that they are incomplete, and do not offer any concrete service to citizens.Even though some informational sites have "Services municipaux" displayed in their menus, which infers the presence of administrative information, they deliver limited services to citizens.Indeed, the pages only provide an inventory of the town's assets, of the institutions that may appeal to tourists, and a services organization chart.Consequently, the level of interactivity of Quebec towns' Web sites appears higher than in French towns' sites.

Discussion
Following the development of the new classification matrix, and then its experimentation within a specific context to test its efficiency, we can underline some difficulties: the first difficulty was to clearly identify a teleprocedure from a mere mail located on a preformatted page (choice of the person to be contacted, objective of the query… from a drop-down menu), and to be sent back to the municipal representatives.Teleprocedures are much more than mere complaint forms sent over some department managers, they come to enrich municipal databases on specific themes: driving licence applications, sending of digital curriculum vitae… where the user has to make choices out of various fields.This differentiation made it difficult to determine whether a site was interactive or transactional.Other researches, following the example of Hadaya (2004), should address this item.another difficulty was to make a clear difference between general information and administrative information.In practice, we have come to make the following distinction: administrative information only targets the citizens' or municipal actors' life (Interactivity type), whereas information relative to the general municipal life, which thus could interest non-citizens, fit the Publishing type of sites.it was difficult to classify an 'hyper-ramified' Web site, for certain of the sections could be very simple, whereas some others could be sophisticated.Our approach consisted in classifying each Web site according to the highest level of sophistication that could be found on the site.Consequently, the matrix developed on an encased approach, that is to say following an imbrication of the functions depending on the levels of maturity, considerably helped the classification process in this case.
All in all, we can conclude that the classification matrix we developed, as well as its criteria grid, are efficient to draw an overall relevant picture.Preliminary observations already revealed that municipalities were very much implied in the diffusion of digital maps (e.g.90% of the Web sites of the Province of Quebec diffuse maps).But on the other hand, the classification matrix enabled us to establish much more cautious optimism, showing very different progress achievements from one town to another, and especially a large majority of maps with only a little amount of dynamism.The criteria proved to be very efficient regarding the classification of the various Web sites, which, according to our qualitative research approach (not statistical), tends to prove so far that the matrix has a high level of intern qualitative validity.
Regarding the external validity of the matrix, that is its capacity to draw an overall relevant picture of a specific municipal situation, it proved to be important as well.Indeed, the matrix emphasized an overall view that seems to be revealing of the organizational situation studied.Yet, as far as we are concerned, we do believe that this external qualitative validity needs further investigations to be strongly ascertained.
In the light of our experimentations, one in Canada and another in France, we consider that we have reached our research objective, which was to develop an efficient tool to classify municipal Web sites using cartographic diffusion technologies.Indeed, this objective has been reached considering the limitations inherent to research approach followed.First of all the research qualitative approach does not validate the classification tool on a statistical point of view.Such a quantitative analysis remains to be carried out.Moreover, it would be interesting to compare the classification approach proposed with other classification tools or means.
Finally, to carry on the present research, we do believe that a literary review and a technology watch may lead to the identification of new classification criteria.This would enrich our classification matrix.Moreover, it would be of interest to test the relevancy of this matrix not only with other types of organizations, but also with other countries.Some longitudinal studies would also be of importance to better understand the municipalities' motivations, and what they expect from their urban geospatial Web sites.Such studies could also confirm the principle according to which a municipality's Web site goes through various stages of development.Indeed, although this "maturity" principle is generally admitted, we think that it should be empirically ascertained in municipal context.

Figure 3 :
Figure 3: Municipal Web sites classification axis and discriminating criteria

Figure 6 :
Figure 6: Online maps classification axis and discriminating criteria

Figure 7 :
Figure 7: GIS-based Web sites classification matrix

Figure 8 :
Figure 8: Progression towards more sophisticated GIS-based Web sites

Figure 9 :
Figure 9: Distribution of the cities analysed according to the types

Figure 10 :
Figure 10: Distribution of the French cities' Web sites within the classification matrix , at best, zoom and pan functions.Yet, if other display functions enrich the cartographic consultation Web interface, the map will match one of the three other types.