NOVEMBER 2000JOURNAL OF TRAVEL RESEARCH

Discovering Next Generation TourismInformation Systems: A Tour on TIScover

BIRGIT PRÖLL AND WERNER RETSCHITZEGGER

Information technology has been playing a fundamentalrole in the tourism industry for a long time. The Internet par-ticularly is increasingly emerging as a perfect platform tobring tourism products direct to the customer. Web-basedtourism information systems, however, are required not onlyto offer online brochures but also to provide both value andservice. The Austrian tourism information system, TIScover,meets this challenge by focusing on three crucial points.First, high-quality access is provided, both by supporting e-commerce transactions and by allowing access to the systemvia the World Wide Web, info kiosks, and cellular phones.Second, the content is of a high quality that is achieved by en-abling tourism information providers to maintain their prod-ucts directly, as well as by integrating external tourism infor-mation sources. Finally, the system is designed in such a waythat it is easy to customize. This facilitates employment of thesystem by the various kinds of tourism information providers,as well as by different regions and even countries.

The tourism industry can be seen as one of the first areasto do business electronically. In the 1960s, airline reservationsystems, which can be regarded as the forerunners of today’se-commerce systems (Copeland and McKenney 1988;Kappel et al. 1996; Kapsammer, Retschitzegger, and Wagner1998; Schulz 1996), were already being employed. Today,tourism information systems are one of the most importantareas of application for e-commerce. The number of touristswho use Web-based tourism information systems for pretripplanning jumped from 3.1 million in 1996 to 33.8 million in1998, a 1,000% increase in 2 years (Travel Industry Associa-tion of America [TIA] 1999b). Furthermore, it is predictedthat within the next 10 years, 30% of the whole tourism busi-ness will be done via the Internet (Schuster 1998). Accordingto a recent report from Gartner Group, the global Internettravel market will increase sixfold by the end of 2001, risingfrom U.S.$5 billion spent at Web sites in 1999 to U.S.$30 bil-lion (Gartner Group Inc. 2000).

There are two major factors, among others, in these promis-ing figures. First, the special properties of tourism products—heterogeneity, intangibility, and perishability—make themvery information intensive and thus especially suited for offer-ing and selling electronically (Sheldon 1997; Werthner andKlein 1999). Second, information technology (IT) will alwaysbe able to cope with the changing demand of tourists towardmore frequent, but shorter travel, last-minute reservations, qual-ity of information and service, market transparency, and tai-lored holidays (Kärcher 1997; Werthner and Klein 1999).Despite this fundamental role of IT in the tourism industry,

when developing tourism information systems today, onehas to keep in mind that only publishing online brochures isby no means enough. Gartner analysts commented that travelcompanies that do not have competitive and easy-to-useInternet sites established by Q2 2000 will be forced to closeoperations or be acquired by stronger online travel firms(Gartner Group Inc. 2000). Three crucial aspects can be iden-tified that seem to be essential for the success of tourisminformation systems; these are quality of access, quality ofcontent, and ability to customize the whole system.

First, the acceptance of a tourism information system par-ticularly depends on comfortable and powerful access capa-bilities. To succeed in this, one prerequisite is that the systemsupports all phases of an e-commerce transaction, from theinformation phase via the negotiation phase to the settlementphase (Kappel, Retschitzegger, and Schröder 1998; Malone,Yates, and Benjamin 1987; Schmid and Lindemann 1998).The information phase especially requires detailed consider-ation to provide the tourist with powerful navigation andsearch capabilities so the desired tourism information orproduct can be found conveniently and efficiently. Anotherprerequisite is that access to the system should not only bepossible in advance of a trip by means of the traditionalWorld Wide Web (WWW). Rather, additional media shouldbe provided to allow the tourists to access tourism informa-tion and products on arrival at the destination or to providemobile access, that is, access independent of location andneed for a laptop.

Second, since information has been identified as one ofthe most important quality parameters for efficient service(Schertler 1994), a tourism information system has to focuson the content aspect by providing comprehensive, detailed,

Birgit Pröll studied computer science at the Johannes KeplerUniversity, Linz, Austria. She has been at the FAW (Research Insti-tute for applied Knowledge Processing) at the University of Linzsince 1991. She has been engaged in industrial and research pro-jects in expert systems and CAD, configuration management, rela-tional and object-oriented databases, and information systems ande-commerce on the Web. Her current research interests are infor-mation retrieval and combining databases and the Web. WernerRetschitzegger received his M.Sc. and Ph.D. from the JohannesKepler University, Linz, Austria. He has been in the Department ofInformation Systems at the Johannes Kepler University, Linz, Aus-tria since 1993. His research interests concern the modeling ofWeb-based and object-oriented systems, the integration ofsemistructured data with object-relational technology, and the ap-plication of Internet and database technology to the area of tourisminformation systems.

Journal of Travel Research, Vol. 39, November 2000, 182-191© 2000 Sage Publications, Inc.

accurate, and topical tourism information. The dynamic natureof some tourism information, for example, rates, schedules,events, opening hours, and weather forecasts changing withthe days, weeks, months and seasons, makes this a challeng-ing task. One possibility to achieve a high-quality content isto allow its maintenance directly by the tourism informationproviders. However, it is not feasible to store all the informa-tion a tourist might be interested in at one Web site, neither interms of storage costs nor, more important, in terms of main-tenance overheads. Another way to provide comprehensivecontents is to use the broad spectrum of tourism informationalready distributed over various Web sites by supportinginteroperability. Integrating this information automaticallywould enable the tourist to collect all tourism information atone place, no matter what is desired, be it information aboutweather and traffic conditions, schedules of trains, planesand buses, or current events.

The third crucial point is that a tourism information sys-tem should represent a generic platform allowing the presen-tation of arbitrary destinations, as well as integrating differ-ent kinds of tourism information providers. One problem inthis respect is the complexity and heterogeneity of tourisminformation and products depending on both the kind of pro-vider and the destination being presented (Malone, Yates,and Benjamin 1987). Another problem is that tourism infor-mation providers, regions, and countries should differ fromeach other, since especially in the tourism industry, unique-ness is the main attraction. Consequently, the system shouldbe easy to customize with respect to different tourism infor-mation providers, regions, and countries while preserving theindividuality of their presentation.

A system that has faced these crucial points is the Aus-trian tourism information system, TIScover. The aim ofTIScover is twofold (Pröll, Retschitzegger, Wagner, andEbner 1998; Pröll, Retschitzegger, and Wagner 1998c): first,tourists should be supplied with comprehensive, accurate,and up-to-date tourism information on countries, regions, vil-lages, and all destination facilities they offer, such as hotels,museums, or other sights. Second, it aims to attract the touristto buy certain tourism products either offline or, if possible,online. TIScover was developed in 1996 based on the experi-ences of the pioneering systems TIS (Ebner 1994) andTIS@WEB (Burger et al. 1997). Originally, TIScover wasdeveloped to market the facilities of just one region of Aus-tria, namely, Tyrol. Since then, six other Austrian regionshave joined TIScover (TIS GmbH 2000a). Moreover, TIScoverhas been employed in Asia, presenting tourism informationabout Thailand (GoThailand 1998)1; it is used by the Germancompany START Media Plus, a major player in the area ofonline reservation systems, to present tourism informationabout Germany (START Media Plus 2000); and it is onlinein Switzerland (TIS GmbH 2000c). Regarding the top 100most frequently accessed travel sites, TIScover Austria wasalready ranked in the top 10, above international companiessuch as American Airlines, Hyatt, Hertz, or American Express(100hot Travel 1999).

This article focuses on the aforementioned crucial aspectsof tourism information systems as they are implemented inTIScover. For this, first, an overview of the whole architec-ture of TIScover, including a short description of the mannerin which the system may be customized, is given. E-com-merce transactions in the context of TIScover are discussed,and different access paradigms supporting the information

phase are emphasized. Afterward, the article deals with mobileaccess to TIScover by means of cellular phones and presentsmechanisms used in TIScover to achieve interoperability. Thearticle concludes with the outlook for future work.

AN OVERVIEW OF TISCOVER

The basic architecture of TIScover consists of three lay-ers, namely, the Public Access Layer, the Internal Layer, andthe External Data Source Layer (see Figure 1).

Public Access Layer

TIScover can be accessed by the public in three ways.First, it can be accessed via the WWW, where the customer isable to navigate through a geographical hierarchy, represent-ing various kinds of tourism information; use different searchfacilities; and even book tourism products online (Pröll,Retschitzegger, Kroiß, et al. 1998). Currently, per month,TIScover Austria has to handle up to 11.5 million pageviews,2

2.6 million visits,3 and as many as 40,000 requests for infor-mation and online bookings. Besides access via the WWW,which is discussed in more detail in the next section, the tour-ist is able to access the tourism information and products of acertain destination by means of info kiosks at the holidaydestination. The functionality of the info kiosk, which is sim-ilar to that of TIScover’s WWW component, can be used viatouchscreen or speech input. Last but not least, TIScover canbe accessed at any time, from anywhere, simply via cellularphones. With this, the tourist is able to get topical informa-tion about weather forecasts, avalanche conditions, and snowreports.

Internal Layer

The Internal Layer is situated beneath the Public AccessLayer. Its core component is represented by the TIScoverdatabase. The database schema of TIScover Austria consistsof approximately 300 database tables and has been con-structed on the basis of a domain data model. This incorpo-rates all conceptual entities gathered during the process ofrequirement definition from numerous tourism informationproviders and from the experiences with the predecessor sys-tems TIS (Ebner 1994) and TIS@WEB (Burger et al. 1997).Currently, the database of TIScover Austria comprises approx-imately two gigabytes of data covering 2,000 towns and vil-lages and nearly 40,000 lodgings. To facilitate efficient andreliable access, Web pages are automatically generated fromthe database every time the underlying data changes. As aresult, more than 400,000 Web pages are stored in some mil-lion files (Pröll et al. 1999). On average, TIScover has to gen-erate 24,000 Web pages per day, representing more than 5%of the whole TIScover Web site. There are three main rea-sons for this high workload: first there are a lot of tourisminformation providers that are able to maintain their data bymeans of the extranet; second, a huge amount of data isalready managed by TIScover; and third and probably mostimportant, the nature of some tourism information is extremelydynamic.

Two other system components placed on top of the TIS-cover database, namely, the extranet and the intranet, providethe basis for a generic platform, allowing the system to be cus-

JOURNAL OF TRAVEL RESEARCH 183

tomized (Pröll, Retschitzegger, and Wagner 1998a, 1998b).The extranet provides authorized tourism information pro-viders, offering all manner of concerns, from small guest-houses to large local tourist offices direct access to their tour-ism information and products (Pröll, Retschitzegger, andWagner 1998a). A particular benefit of tourism informationproviders is that they are allowed to maintain their tourisminformation and products decentrally and customize them invarious ways. For example, a tourism information provider isable to decide the number and complexity of the productspresented, ranging from single products to whole packages(Kärcher 1997; Pröll, Retschitzegger, and Wagner 1999) andto customize their presentation, for example, by choosingbetween different layout templates, thus coping with thedemand for individuality. At this time, the extranet ofTIScover Austria is accessible to 7,000 tourism informationproviders. In contrast to the extranet, the intranet componentof TIScover is accessible to the system provider only andallows the whole system to be customized. It is, for example,possible to extend the geographical hierarchy to integratenew countries or villages into the system, to define defaultlanguage and currency for all system components, and toconstruct layout templates that can then be selected by thetourism information provider via the extranet. The variouspossibilities to customize the system as offered by theextranet and the intranet have proven to be very powerful. Noprogramming effort was needed to customize the system forThailand, Germany, and Switzerland, three significantly dif-ferent countries where the nature of their tourism facilities isconcerned.

External Data Source Layer

Although TIScover manages a huge amount of tourisminformation, it is of course far from being complete. TheExternal Data Source Layer ensures interoperability to otherWeb sites going beyond simple links, to satisfy requests forcertain information that is not part of the TIScover databasebut is already available at other Web sites. The External DataSource Layer has been prototypically realized in the courseof the ESPRIT project MIRO-Web to federate multiplestructured and semistructured tourism information sourceson the Web (Haller et al. 2000).

E-COMMERCE TRANSACTIONS IN TISCOVER

Today’s tourism information systems are required tofully support an e-commerce transaction. The interactionprocesses between the participants of an e-commerce trans-action can be grouped into three phases: the informationphase, the negotiation phase, and the settlement phase (Schmidand Lindemann 1998). In the following section, the conceptsand mechanism employed by TIScover to support thesephases are described in more detail.

The Information Phase

In the information phase, customers, that is, tourists, ac-quire a market overview by gathering information about po-tential market partners: tourism information providers, as well

184 NOVEMBER 2000

ExtranetTourism InformationProvider

WWW

IntranetSystem Provider

TIScover Database

Customizing

Customizing Customizing

Tourist

Cellular PhoneInfo Kiosk

PublicAccessLayer

InternalLayer

ExternalData SourceLayer IDTrgTableIDTrgTable

......

......

........................

XML Doc.<a> <b>...</b> <c d=.../></a>

IDTrgTableIDTrgTable

......

......

........................

.... ......... ...

.... ......... ...

.... ......... ...

.... ......... ...

IDTrgTableIDTrgTable..................

......

......

......

XML Doc.<a> <b>...</b> <c d=.../></a>

IDTrgTableIDTrgTable

......

......

........................

.... ......... ...

.... ......... ...

.... ......... ...

.... ......... ...

FIGURE 1BASIC ARCHITECTURE OF TISCOVER

as their products and services (Fröschl and Werthner 1997).To provide a comfortable information access, TIScover sup-ports three paradigms for finding tourism information andproducts (see Figure 2).

• First is the hypertext paradigm (Conklin 1997) for nav-igating through information by using hypertext links.This is realized by means of the Atlas module.

• Second, the database query paradigm for a preciselystructured search (Domenig and Dittrich 1999) on thebasis of a common database schema is provided by theScout module.

• Third is the classical information retrieval paradigm(Salton and McGill 1983), providing a full text searchas it is done by most search engines on the Web.

The Atlas module allows the customer to browse throughall kinds of tourism information by navigating through a geo-graphical hierarchy ranging from world via continent, state,and country to a village. Each geographical level offers thosepieces of tourism information and products that are impor-

tant at this level. For example, a tourist may get informationabout duty free regulations or highlights of a state at state level,whereas at the village level, information about skiing facili-ties is available. Both the kind of information and the contentgets more detailed the deeper one navigates into the hierar-chy. Accommodations—for example, hotels—are placed atthe village level and represent the starting point for onlinebooking (cf. Section 3.3). To prevent tourists from followinglinks with products that cannot be booked online, the book-ing button is only visible if there are products within a villageand at least one of them is available for online booking. Fur-thermore, the Atlas does not have to access the database dur-ing navigation but can work fully on Web pages, which as al-ready mentioned, are generated from the database every timethe underlying data changes (Pröll et al. 1999).

The Scout module represents the second way offered byTIScover to find tourism information and products. In con-trast to Atlas, Scout makes navigation completely superflu-ous in that it realizes a precisely structured search within theTIScover database by allowing the specification of varioussearch criteria, which are represented by attributes of the

JOURNAL OF TRAVEL RESEARCH 185

Precise Structured Search Navigation

Index

Entry of Arbitrary Search Criteria

Ranked Result List

TIScover Database

Full Text Search

Web Pages

State

Country

Village/City

Accommo -dation

Product

Continent

WWWTourist

......

......

........................

.... ......... ...

.... ......... ...

.... ......... ...

.... ......... .....................

......

......

......

.... ......... ...

.... ......... ...

.... ......... ...

.... ......... ...

Unordered Result List

Entry of Predefined Search Criteria

FIGURE 2ACCESS PARADIGMS SUPPORTING THE INFORMATION PHASE

common database schema. These criteria include location,name, and facilities of the accommodation, type of room,type of board, price limits, and arrival and departure dates. Ingeneral, the Scout module is especially appropriate for thesingle-minded tourist who already has a clear idea of what heor she wants. A major benefit of Scout is that products, whichcan be booked online, can be searched at any geographicallevel and the system supplies only those products available atthe required time. Note that this is different from Atlas,where the tourist has to navigate down to the lowest geo-graphical level, namely, villages, before there is any possibil-ity to check the availability of products.

Finally, TIScover allows a tourist to start a full textsearch, at any geographical level, to find all Web pages con-taining an arbitrary keyword, for example, all pages contain-ing the keyword “carving” starting at the level of Tyrol. Thesystem matches the keyword against an index built by arobot. The robot is started from time to time to index Webpages. This approach is comparable to well-known Internetsearch engines, stemming from traditional informationretrieval (Salton and McGill 1983). One advantage of thisapproach is that the customer is free to choose the searchkeywords. This is in contrast to a structured search in whichthe customer is restricted to those keywords that are prede-fined by the system provider and are, thus, part of the com-mon database schema. Another advantage is that unlike thestructured search, in which the result set of a query can onlybe empty or non-empty, the results of a full text search areranked (Salton and McGill 1983). However, there are alsosome limitations. First, the tourist has no guidance in usingthe right keywords; thus, it is quite hard to find words match-ing the index of the Web pages. For example, one will obvi-ously not succeed by using synonyms that are not part of theindex. Second, there is a considerable system overhead everytime the robot is started to index the pages again. And third,the less often the robot is started, the higher the probabilitythat the search is processed on an index that is outdated.

It is important to note that the three access paradigmsimplemented by TIScover are not isolated from each other.Instead, they are seamlessly integrated so that a tourist canarbitrarily switch in-between and the system automaticallypreserves all necessary context information. When, for exam-ple, browsing on the Tyrolean level of Atlas, a switch toScout or to full text search implies that the following searchis done automatically in the context of Tyrol only. If the tour-ist navigates from the result list of a search for products onAtlas, the context for browsing is automatically the villagethe product is related to.

The Negotiation Phase

In the negotiation phase, the conditions of the e-commercetransaction are negotiated. The conditions for tourism prod-ucts are, for example, facilities at the accommodation, typeof board, duration of stay, price, and terms of payment. In thesimplest case, these conditions are predefined in full, and thecustomer cannot tailor a certain product to his or her individ-ual needs. A more sophisticated negotiation phase would bebased on some negotiation protocol, in which one or bothparticipant(s) are able to negotiate the final terms.

In TIScover, the tourism information provider is respon-sible for defining the negotiation protocol for tourism prod-ucts in that he or she specifies the conditions according to

which he or she is willing to sell them. There can be productsthat are fully predefined (e.g., a last-minute offer) but alsoproducts where the tourism information provider specifiesthe product’s conditions but the customer is able to tailor theproduct to his individual needs. TIScover allows the sametourism product to be offered with different conditions. Forexample, different types of board or different seasons imply aprice reduction or increase. With time conditions, it can bespecified that a tourism product is only available during acertain season and valid on weekdays only. Finally, it can beascertained whether credit cards are accepted and whichones, and whether payment by bank transfer is possible.

If a customer searches for a tourism product via the Scoutmodule, the information phase and the negotiation phaseoverlap, since to carry out a search, the customer has todefine the criteria according to which he is willing to book aproduct. Depending on how precisely these criteria havebeen specified and whether the product has been fully prede-fined by the tourism information provider, the results listcontains products that allow no further customization andcan be booked directly or products that can be further adaptedto individual needs, for example, by choosing the type ofboard (see Figure 3). It has to be emphasized that each modi-fication of the conditions implies an immediate recalculationof the price.

The Settlement Phase

In the settlement phase, both parties fulfil the termsagreed in the negotiation phase. As is the case with manyproducts sold electronically, only part of the settlement phasecan be supported by the system for tourism products sincethere is a difference between the time of booking and the timeof consuming. Existing tourism information systems oftendo not really support the settlement phase. Some offer only atelephone number, others offer only an e-mail address, with-out checking availability or dealing with a proper actualiza-tion of the inventory.

To book products in TIScover, tourists have to enter theirpersonal data; their credit card numbers, if payment by creditcard has been chosen; and confirmation that they accept thebooking conditions. If tourists decide to pay by bank transfer,they get an automatically generated e-mail with the requestto confirm the booking by forwarding this e-mail to theprovider.

The internal booking process comprises several steps.First, the personal data of the tourist and all the data concern-ing the product booked are stored in the database. Second,the inventory of the product is actualized. Third, the tourisminformation provider is notified of the booking by an auto-matically sent fax and e-mail. Finally, the tourist is sent avoucher confirming his or her booking. All data transferredduring the booking process are encoded via the SecureSockets Layer protocol (Netscape Inc. 1999).

MOBILE ACCESS TO TISCOVER

With the rapid growth of wireless telecommunication onone hand and the exponential expansion of the Internet on theother hand, the convergence of these two mainstreams is pre-dictable. Mobile access to Internet content and services bymeans of mobile devices, such as cellular phones, is indeed

186 NOVEMBER 2000

becoming an interesting possibility, both for the tourism infor-mation provider to reach new customers and for tourists onthe move. Mobile services are not replacing pure Web appli-cations or interactive television services (Rose et al. 1999),but they complement them in that for example, topical infor-mation is accessible both at home and wherever is most con-venient for the tourist. The market for this new applicationarea seems to be ready since according to the latest TIAtravel poll of 1,200 U.S. adults, 51% of the 39.8 million busi-ness travelers and 46% of the pleasure travelers over the pastyear said they took their cellular phone with them on a trip(TIA 1999a). Furthermore, according to the Strategis Group(1999), there will be more than 530 million owners of cellu-lar phones by the year 2001, and a substantial proportion ofthe phones will allow Internet access.

To stay abreast of these promising changes, TIScoveralready provides three different mobile services, namely,weather forecasts, avalanche conditions, and snow reports(TIS GmbH 2000b). This kind of tourism information is alsoavailable via the traditional Web, but it is especially wellsuited to mobile access. This is because it represents a hotspot within the TIScover database, being subject to a particu-larly high update frequency. At the same time, the value ofthis information depends on its topicality and, therefore, onthe ability of the tourist to access this information at anytime, independent of location. In the following section, the

basic technology for realizing mobile access and its usage inTIScover is described in more detail.

W@P—The Basic Technologyfor Realizing Mobile Access

The basic technology for enabling mobile access andrealizing mobile services is the Wireless Application Proto-col (W@P) (Wireless Application Forum Ltd. 1999). W@Pis the de facto world standard for the presentation and deliv-ery of information and telephone services on cellular phonesand other mobile devices. W@P belongs to an industry con-sortium of approximately 200 members, including 75% ofthe mobile operators and nearly all phone manufacturers.The specification of W@P includes three basic buildingblocks (see Figure 4).

First, the Wireless Markup Language (WML), a subset ofthe Extensible Markup Language (XML) (World Wide WebConsortium [W3C] 2000) and the Hypertext Markup Lan-guage (HTML) (W3C 2000), is used to describe the content,which should be displayed on the mobile device and is espe-cially designed to cope with the limited screen size and inputcapabilities of mobile devices. Second, a microbrowser inthe mobile device controls the user interface, that is, defineshow WML should be interpreted, and is analogous to a stand-ard Web browser so the user can navigate back and forth

JOURNAL OF TRAVEL RESEARCH 187

FIGURE 3CUSTOMIZING TOURISM PRODUCTS

between several WML pages. Finally, to take into accountthe specific characteristics of wireless networks, a W@PGateway, which is usually provided by mobile operator com-panies, is responsible for mediating between the mobiledevice and the server offering the WML pages. ProtocolAdapters transform special W@P protocols to standardHypertext Transfer Protocol (HTTP) requests (W3C 2000)and vice versa. The protocols are defined for use with thelow-bandwidth and high-latency conditions often found inwireless networks. Encoders and Decoders translate textualWML pages into a more compact binary format and viceversa, to reduce the size and number of packets traveling overthe wireless network.

A Typical Scenario:Mobile Access to Snow Reports

Based on the technologies described above, the architec-ture used for enabling mobile access to TIScover is illus-trated in Figure 4. Since W@P is based on standards for Webapplication development such as HTTP and XML/HTML,mobile access to TIScover can be enabled with minimaladditional effort and without changing any existing compo-nent. On the server side, WML documents are generatedfrom the TIScover database and are offered by the TIScoverstandard Web server, in the same way as HTML pages usedfor access via WWW.

To use the mobile services provided by TIScover, thetourist has to connect to TIScover by simply entering therespective unified resource locator (URL) into his or her cel-lular phone. After the connection has been established, theTIScover logo is shown, and the tourist can select one of theservices displayed, that is, snow reports, weather forecasts,or avalanche conditions (see Figure 5). After selecting theservice “snow reports,” a list of all Austrian regions isshown, and the tourist has to select one option. Now, thetourist is able to enter the first letter, but preferably the fullname, of the village or skiing area desired. If the name is mis-spelled, TIScover uses a phonetic search to find names withinthe database that sound similar. If the results list containsmore than one entry, the tourist can scroll up and down andselect one of the entries or can navigate back and startanother search. As soon as an entry has been selected, thesnow report is displayed, comprising name of the village/ski-ing area, date, snow level, and snow quality for mountain,valley, and village. Furthermore, direct links to both weatherforecasts and avalanche conditions of this geographical areaare provided.

INTEROPERABILITY OF TISCOVER

Although quality of access is a major prerequisite for asuccessful tourism information system, as already mentioned,quality of content is equally, if not more, important. To fulfillthis requirement, it is of course infeasible to store all infor-mation a tourist might be interested in at one Web site, bothin terms of storage costs and, more important, in terms ofmaintenance overhead. A more promising idea would be totake advantage of the huge amount of other relevant tourisminformation that is already distributed throughout the Web.However, most of current Web sites do not allow for the inte-gration of multiple data sources beyond simple links betweenthem.

In the course of the ESPRIT project MIRO-Web (Fank-hauser et al. 1998), which was started at the end of 1997,TIScover has been extended with an External Data SourceLayer to federate multiple structured and semistructuredtourism information sources on the Web. These sources com-prise different in-house databases employed in the varioustourism offices, Excel files used for managerial purposes,and finally, HTML pages of other Web sites. In the followingsection, the concepts and technology employed in MIRO-Web are discussed along with a typical usage scenario.

Basic Concepts and Technology of MIRO-Web

MIRO-Web builds on the technology already developedand the knowledge acquired in the course of the EU ESPRITproject IRO-DB aiming at the integration of heterogeneousdatabases (Kapsammer, Retschitzegger, and Wagner 1998).In the spirit of IRO-DB, MIRO-Web is based on a three-tierarchitecture, consisting of a Data Source Adapter Layer, aMediation Layer, and a Client Layer (see Figure 6).

The Data Source Adapter Layer consists of a number ofadapters, also known as wrappers, which are needed to maskthe heterogeneity of data sources and to transform sourcedata into a structured format (Huck et al. 1998). Data sourceadapters have two main functions: first, they translate a queryto the underlying query system used by the source, and then,they translate the results sent back. MIRO-Web adapters canbe used in conjunction with a mediator (cf. below) or independ-

188 NOVEMBER 2000

Gateway

WML Encoder/ Decoder

Protocol Adapters

Client

WAPProtocols

TIScover Database

HTTP

......

......

........................

WML Doc.<a> <b>...</b> <c d=.../></a>

MicroBrowser

FIGURE 4ARCHITECTURE OF TISCOVER’S MOBILE ACCESS

Accept

Previous

FIGURE 5NAVIGATING THROUGH SNOW REPORTS

ently by an application. MIRO-Web supports several kindsof adapters, relational adapters, structured, and semistruc-tured file adapters, depending on the nature of the datasource. To facilitate both building and maintenance of datasource adapters, a Java toolkit called “Adapter DevelopmentKit” has been developed.

According to (Wiederhold 1992), mediators “simplify,abstract, reduce, merge and explain data.” The MediationLayer of MIRO-Web provides the means to combine andintegrate these heterogeneous sources into a homogeneousview and supports queries about this view in terms of a singlepoint of access. MIRO-Web supports a materialized approachto mediation as well as a virtual approach, whereby it is theresponsibility of the database administrator to decide whichapproach is suitable for a certain data source. In the formercase, data from multiple Web sources are stored within anOracle8 database (Oracle Corporation 2000); whereas, in thelatter case, external data are not accessed before the userissues a query. Virtual mediation is performed by DISCO(Distributed Information Search COmponents), a distributedextensible query engine that provides a uniform query lan-guage and data model for declarative access to a heteroge-neous collection of data sources (Tomasic, Raschid, andValduriez 1996).

The Client Layer provides Web-based user interfaces,allowing both the expression of queries to the underlying het-erogeneous sources and the uniform presentation of theirresults.

A Typical Scenario: The Event Agent

To demonstrate the suitability of MIRO-Web to TIScover,an exemplary scenario will now be presented, in which somekind of agent assists the tourist in querying heterogeneous

tourism information sources in a transparent way. A commoncase is that a tourist wants to find a hotel near some eventlocation, such as an exhibition, a sporting event, or a culturalevent. Furthermore, the weather forecast should be good andthe hotel should cost less than a certain amount of money,issued in an arbitrary currency. To specify such a request, theEvent Agent provides a uniform graphical interface in termsof a Java applet (see Figure 7).

The Adapter Layer of MIRO-Web extracts the answer forthis request from different heterogeneous data sources. First,information about hotels is extracted directly from theTIScover database. Second, information about current eventscan be found at different Web sites that already exist (cf., e.g.,www.austria-tourism.at or www.film.at or www.events.at).Third, the weather information is gathered via file transferprotocol in the form of a structured file. And finally, theexchange rate for calculating the currency requested is avail-able in a semistructured format, namely, as an e-mail.

The basic join criteria for the query, which is used by theMediation Layer to satisfy an information request, is thename of the location. On the basis of this join attribute, infor-mation about the events can be joined with weather informa-tion and hotel information. The price can be calculated andconverted by means of the exchange rates and joined with thehotel information. Finally, the Client Layer, that is, the Javaapplet, presents the result of the information request in a uni-form way on the right side of the panel (see Figure 7).

OUTLOOK

Future work is planned with respect to all three areasdescribed in this article, that is, e-commerce transaction sup-port, mobile access, and interoperability.

Where the support of e-commerce transactions is con-cerned, a visionary approach would be to allow the tourist notjust to book accommodation by means of TIScover and con-tact Web sites, for example, of car rental companies or air-lines individually, but rather to allow the tourist to assemble a

JOURNAL OF TRAVEL RESEARCH 189

FIGURE 6OVERALL ARCHITECTURE OF MIRO-WEB

FIGURE 7EVENT AGENT USER INTERFACE

whole holiday package and to book the package at once(Kappel et al. 1996; Pröll, Retschitzegger, and Wagner1999). Two scenarios of varying complexity can be distin-guished. First, the Web sites, whose products can be selectedto assemble the package, could be predefined. This is similarto electronic shopping malls where the shops participatingare known in advance. Second and more visionary would befor the user to navigate from TIScover to arbitrary Web sitesoffering products, and some runtime engine checks whetherthe Web site is capable of participating in a global bookingprocess (Kappel, Rausch-Schott, and Retschitzegger 1999).

Mobile access poses several interesting possibilities forfuture developments. With respect to e-commerce transac-tions, W@P provides the proper technology to allow secureonline transactions. This means that it will be possible in thenear future for a tourist to book tourism products anywherevia cellular phones. Furthermore, W@P will define a “pushmechanism,” enabling any Web server to send information tothe client, that is, the cellular phone, without having to waitfor information requests. This function would alert subscrib-ers when time-sensitive information changes. For example,immediate information about a predicted fall in temperaturewould be of great value for an alpinist. Finally, W@P alsoprovides location information, meaning that the mobile ser-vices can vary depending on where one uses them. An inter-esting application scenario would be a location-dependenttourist guide for city visitors, where the visitors are providedwith context-sensitive information, such as where to find thebest restaurant, when the next bus leaves from that bus stop,or what the current waiting times at local attractions are(Davies et al. 1998; Hartl et al. 2000).

Concerning interoperability, future work will be doneprimarily in the course of the follow-up EU ESPRIT projectof MIRO-Web, called XML-KM (XML-based Mediator forKnowledge Extraction and Brokering). The goal of XML-KM, which is strongly based on XML, is to enhance the com-ponents and tools developed in the course of MIRO-Web tobe able to collect and disseminate knowledge instead of justdata. Through a rule-based XML-wrapper, tourism informa-tion from corporate databases, HTML files, and office appli-cations will be collected in data warehouses (Gardarin, Sha,and Ngoc 1999). Using classical data mining tools, knowl-edge will be extracted from the warehouses in the form ofderived tourism information and rules. Through XML-basedquery tools, users will be able to subscribe and receive per-sonalized tourism information in an appropriate format onvarious devices including computers, cellular phones, andfaxes.

NOTES

1. Note that due to the economic crisis in Asia, TIScoverasia iscurrently not operating.

2. Pageviews depict the number of “real” documents transmit-ted by a Web server without counting access to images, CGI scripts,or Java applets.

3. Visits count the number of users accessing one or more pagesof a Web server.

REFERENCES

Burger, F., P. Kroiß, B. Pröll, R. Richtsfeld, H. Sighart, and H. Starck(1997). “TIS@WEB—Database Supported Tourist Information onthe Web.” In Proceedings of the International Conference on Informa-tion and Communication Technologies in Tourism (ENTER), edited byA. Min Tjoa. Edinburgh, Scotland: Springer, pp. 180-89.

Conklin, J. (1997). “Hypertext: An Introduction and Survey.” IEEE Com-puter, 20 (9): 17-41.

Copeland, D., and J. L. McKenney (1988). “Airline Reservation Systems:Lessons from History.” MIS Quarterly, 353-70.

Davies, N., K. Mitchell, K. Cheverst, and G. Blair (1998). “Developing aContext Sensitive Tourist Guide.” In Proceedings of the First Work-shop on Human Computer Interaction with Mobile Devices [Online].University of Glasgow, Scotland. Available: http://www.dcs.gla.ac.uk/~johnson/papers/mobile/HCIMD1.html#_Toc420 818986

Domenig, R., and K. R. Dittrich (1999). “An Overview and Classification ofMediated Query Systems.” SIGMOD Record, 28 (3): 63-72.

Ebner, A. (1994). “TIS—Tirol Informations System: Die Konsequenz einerIdee.” In Tourismus, edited by W. Schertler, Ueberreuter. In German,pp. 307-45.

Fankhauser, P., G. Gardarin, M. Lopez, J. Munoz, and A. Tomasic (1998).“Experiences in Federating Databases: From IRO-DB to MIRO-Web.” In Proceedings of the 24th International Conference on VeryLarge Data Bases (VLDB), edited by A. Gupta et al. New York: Mor-gan Kaufmann, pp. 655-58.

Fröschl, K. A., and H. Werthner (1997). “Informed Decision Making inTourism Management—Closing the Information Circuit.” In Pro-ceedings of the International Conference on Information and Commu-nication Technologies in Tourism (ENTER), edited by A. Min Tjoa.Edinburgh, Scotland: Springer, pp. 75-84.

Gardarin, G., F. Sha, and T. D. Ngoc (1999). “XML-based Components forFederating Multiple Heterogeneous Data Sources.” In Proceedings ofthe 18th International Conference on Conceptual Modeling (ER), ed-ited by J. Akoka et al., LNCS 1728. Paris: Springer, pp. 506-19.

Gartner Group Inc. (2000). GartnerGroup’s Online Travel Forecast: $30Billion by 2001 [Online]. Available: http://gartner12.gartnerweb.com/public/static/aboutgg/pressrel/pr010600.html

GoThailand (1998). TIScover ASIA [Online]. Available: http://www.tiscoverasia.com

Haller, M., B. Pröll, W. Retschitzegger, A. M. Tjoa, and R. R. Wagner(2000). “Integrating Heterogeneous Tourism Information in TIScover—The MIRO-Web Approach.” In Proceedings of the International Con-ference on Information and Communication Technologies in Tourism(ENTER), edited by D. R. Fesenmaier et al. Barcelona, Spain: Springer,pp. 71-80.

Hartl, A., G. Austaller, G. Kappel, C. Lechleitner, M. Mühlhäuser, S. Reich,and R. Rudisch (2000). “Gulliver—A Development Environment forWAP Based Applications.” In Poster Proceedings of the 9th Interna-tional World Wide Web Conference, Amsterdam, the Netherlands,pp. 26-7.

Huck, G., P. Fankhauser, K. Aberer, and E. Neuhold (1998). “Jedi: Ex-tracting and Synthesizing Information from the Web.” In Proceedingsof the 3rd IFCIS International Conference on Cooperative Informa-tion Systems (CoopIS), IEEE Computer Society Press, pp. 32-43.

Kappel G., S. Rausch-Schott, and W. Retschitzegger (1999). “TransactionSupport for DataWeb Applications—A Requirement’s Perspective.”In Proceedings of the Fifth Americas Conference on Information Sys-tems (AMCIS), Milwaukee, WI, pp. 877-79.

Kappel, G., S. Rausch-Schott, W. Retschitzegger, and M. Sakkinen (1996).“A Transaction Model for Handling Composite Events.” In Proceed-ings of the 3rd International Workshop of the Moscow ACM SIGMODChapter on Advances in Databases and Information Systems (ADBIS),Moscow, Russia, pp. 116-25.

Kappel, G., W. Retschitzegger, and B. Schröder (1998). “Enabling Technol-ogies for Electronic Commerce.” In Proceedings of the XV. IFIP WorldComputer Congress, Vienna, Austria/Budapest, Hungary, pp. 396-400.

Kapsammer, E., W. Retschitzegger, and R. R. Wagner (1998). “MetaData-Based Middleware for Integrating Information Systems: A CaseStudy.” In Proceedings of the 4th International Conference on Infor-mation Systems Analysis and Synthesis (ISAS), Orlando, FL, pp. 177-83.

Kärcher, K. (1997). Reinventing the Package Holiday Business. Wiesbaden,Germany: DUV.

Malone, T. W., J. Yates, and R. I. Benjamin (1987). “Electronic Markets andElectronic Hierarchies.” Communications of the ACM (CACM), 30(6): 484-97.

Netscape Inc. (1999). SSL 3.0 Specification [Online]. Available: http://www.netscape.com/eng/ssl3/index.html

100hot Travel (1999). The Web’s Most Popular Travel Sites [Online].Available: http://www.100hot.com/travel/

190 NOVEMBER 2000

Oracle Corporation (2000). http://www.oracle.com/Pröll, B., W. Retschitzegger, P. Kroiß, and R. R. Wagner (1998). “Online

Booking on the Net—Problems, Issues and Solutions.” In Proceedingsof the 5th International Conference on Information Technology inTourism (ENTER), edited by D. Buhalis et al. New York: Springer,pp. 268-77.

Pröll, B., W. Retschitzegger, H. Sighart, and H. Starck (1999). “Ready forPrime Time—Pre-Generation of Web Pages in TIScover.” In Pro-ceedings of the 8th International ACM Conference on Information andKnowledge Management (CIKM), Kansas City, MI, pp. 63-8.

Pröll, B., W. Retschitzegger, and R. R. Wagner (1998a). “Extranet-BasedMaintenance and Customisation of Tourism Information.” Technol-ogies for the Information Society: Developments and Opportunities,edited by Roger, J. -Y. et al. Burke, VA: IOS Press, pp. 331-38.

(1998b). “TIScover—A Tourism Information System Based onExtranet and Intranet Technology.” In Proceedings of the 4th Ameri-cas Conference on Information Systems (AIS), Baltimore, pp. 1014-16.

(1998c). “TIScover—A Web-Based Tourist Information System.”In Proceedings of the EURO-MED NET Conference, Nicosia, Cyprus,Greece, pp. 93-4.

(1999). “Holiday Packages on the Web.” In Proceedings of the In-ternational Conference on Information and Communication Technol-ogies in Tourism (ENTER), edited by D. Buhalis et al. Innsbruck,Austria: Springer, pp. 108-18.

Pröll, B., W. Retschitzegger, R. R. Wagner, and A. Ebner (1998). “BeyondTraditional Tourism Information Systems—TIScover.” Journal of In-formation Technology in Tourism (ITT), Inaugural Volume, CognizantCorp., USA, pp. 15-31.

Rose, M., C. Dormann, H. Olesen, B. Beute, and J. F. Jensen (1999). “WhitePaper on Interactive TV.” Interim report of the InterMedia project ofthe Danish National Center for IT-Research [Online]. Denmark. Avail-able: http://www.cti.dtu.dk/personal/berco/html/cti/ITV%20white%20paper.html.

Salton, G., and M. McGill (1983). Introduction to Modern Information Re-trieval. New York: McGraw-Hill.

Schertler, W. (1994). “Tourism 2000—An Information Business.” In Pro-ceedings of the International Conference on Information and Commu-nication Technologies in Tourism (ENTER), edited by W. Schertler etal. Vienna/New York: Springer, pp. 20-6.

Schmid, B. F., and M. A. Lindemann (1998). “Elements of a ReferenceModel for Electronic Markets.” In Proceedings of the 31st Annual Ha-waii International Conference on Systems Science (HICSS), vol. IV.Los Alamitos, CA: IEEE Computer Society Press, pp. 193-201.

Schulz, A. (1996). “The Role of Global Computer Reservation Systems inthe Travel Industry Today and in the Future.” Newsletter CompetenceCenter Electronic Markets [Online], 6 (2). University of St. Gallen.Available: http://www.businessmedia.net/netacademy/publications.nsf/all_pk/77

Schuster, A. G. (1998). “A Delphi Survey on Electronic Distribution Chan-nels for Intermediaries in the Tourism Industry: The Situation in Ger-man Speaking Countries.” In Proceedings of the International Confer-ence on Information and Communication Technologies in Tourism(ENTER), edited by D. Buhalis et al. Vienna/New York: Springer,pp. 224-34.

Sheldon, P. J. (1997). Tourism Information Technology. New York: CabInternational.

START Media Plus (2000). TIScover Germany [Online]. Available: http://www.deutschlandreise.de

Strategis Group (1999). CellTRAC: Cellular and PCS Consumer Trends:1999—(Latest Update) [Online]. Available: http://www.strategisgroup.com/

TIS GmbH (2000a). TIScover Austria [Online]. Available: http://www.tiscover.com

(2000b). TIScover Austria W@P Service [Online]. Available: http://wap.tiscover.com

(2000c). TIScover Switzerland [Online]. Available: http://www.tiscover.ch

Tomasic, A., L. Raschid, and P. Valduriez (1996). “Scaling HeterogeneousDatabases and the Design of DISCO.” In Proceedings of the 16th In-ternational Conference on Distributed Computing Systems. HongKong: IEEE Computer Society, pp. 449-57.

Travel Industry Association of America (1999a). Americans Are “StayingWired” While Traveling [Online]. Available: http://www.tia.org/press/091999poll.stm

(1999b). New Travel and Technology Report: 6.7 Million AdultsUse the Internet For Travel Reservations [Online]. Available:http://www.tia.org/press/010899tech.stm

Werthner, H., and S. Klein (1999). Information Technology and Tourism—AChallenging Relationship. Vienna, Austria/New York: Springer.

Wiederhold, G. (1992). “Mediators in the Architecture of Future Informa-tion Systems.” Computer, 25 (3): 38-49.

Wireless Application Forum Ltd. (1999). “Wireless Application Protocol(WAP)” [Online]. White Paper. Available: http://www.wapforum.org

World Wide Web Consortium (2000). Available: http://www.w3.org/

JOURNAL OF TRAVEL RESEARCH 191