I dialog med brugerne gennemføres en videreudvikling af By og Bygs website, www.by-og-byg.dk, som primært henvender sig til professionelle praktikere der beskæftiger sig med byggeriet og det byggede miljø. Ex ante antages det at såvel navigationsstruktur som søgefaciliteter skal rekonstrueres. Projektet er stærkt empirisk orienteret og omfatter både undersøgelser af brugerbehov og afprøvning af en prototype. I forlængelse af denne empiri indeholder afhandlingen en række refleksioner over de anvendte metoder. Gennem kombinationen af forskellige empiriske metoder (metodetriangulering) konstateres bl.a. et behov for at gøre resultater af søgninger på www.by-og-byg.dk mere overbliksdannende. Som svar på dette behov konstrueres en sortering af søgeresultater, som indebærer dels fjernelse af redundante resultater, dels en struktureret adgang til søgeresultaternes indhold. Det vurderes at andre lignende informationstunge websites med fordel vil kunne hente inspiration i den udviklede løsning. Afhandlingen beskriver endvidere en komplet rekonstruktion af By og Bygs website, inkl. en ny navigationsstruktur som imødekommer brugernes behov for en emnebaseret indgang til websitets indhold. Ved den afsluttende brugerafprøvning fremkommer det - noget overraskende - resultat at testpersonerne umiddelbart har vanskeligt ved at identificere websitets hypermedierede sammenfatninger af forskningsresultater, idet testpersonerne på forhånd forventer at By og Bygs forskningsresultater udelukkende foreligger i forskningsrapporter i traditionel, lineær form. Ex post synes det således fortsat at være en stor udfordring at hypermediere forskningsresultater med henblik på anvendelsesorientering. Afhandlingen ledsages af en betaversion af By og Bygs ny website, som blev idriftsat april 2004 på adressen www.by-og-byg.dk. Danish Building and Urban Research (DBUR) is a national research institute affiliated to the Danish Ministry of Economics and Business Affairs. The institute aims at providing public and private institutions and enterprises in the construction and housing sectors with research based knowledge, in order to improve the build environment. The subject of this thesis is the reconstruction of DBUR's Danish speaking website on the address www.by-og-byg.dk. The reconstruction is based on a variety of methods for user-dialogue and surveys among the website's target audience of professional practitioners (Argyris & Schön, 1974).FindingsThe pre-assumption that the existing website's navigation structure as well as search functionality need to be changed, is supported by dialogue with a user-panel. The user-panel asks for a navigation structure based on research subjects, and a better representation of search results, and particularly the panel points out a need for reducing the number of links in each search result. Based on focus group interviews (Morgan, 1993) it seems that the general users of DBUR mainly seek information when they are confronted with a problem or a situation in which they need to reduce risk. Only rarely they seek information out of pure general interest. Thus the results of the focus group interviews are in line with common theory on how professionals think when they work (Schön, 1983). Often the interviewed find that DBUR's research is far too distant from practise. Also, the focus-group interviews identify a demand for computermediated access to information resources across the different knowledge institutions in the housing and construction sectors. Finally, the interviews show some differences between the seven interviewed profession groups, perhaps most evidently regarding the concept of knowledge and learning, where particularly one profession (architects) emphasize the importance of aesthetic learning, while the same profession more or less disregard the importance of the written word. Based on a log-file analysis and an on-line questionnaire survey it seems clear that a large majority of users of the existing website is not interested in information on DBUR's organisational matters, as they solely use the website to retrieve information on research results and related, practical recommendations. Testing of a running prototype (J. Nielsen, 1993) of the reconstructed website shows that the invented subject index is clear speech to the six test persons, and that they generally find it easy to navigate the site. The search facility seems to work well, too, although failures in the style sheet distort the test results. Surprisingly, however, the test also shows that the test persons have difficulties in recognizing that the website contains hypertexted (Landow, 1997, ; J. Nielsen, 1990) summaries of research results. It seems as if the presence of such material conflicts with the test persons' expectations and image of DBUR as a governmental research institute that sticks to solid research traditions. Thus, the test persons expect that DBUR's research results can only be found in linear reports (i.e. pdfs), and not in a hypertext format.InventionsThe reconstructed website is based on Intrasuite, which is a standard content management system (CMS), with a modified web-connector. The modifications are technically developed by the software-company Dansk Internet Selskab, instructed by DBUR's communications department. The reconstructed website features a special way in which to present search results from the standard Microsoft search engine applied on the website. The problem is that a full-text search often returns several links to inter-related contents, which reduces the usability of the search result. This problem is tackled by the invention described as follows: While the search procedure is still based on full-text-search, the presentation of search results is generated by the use a special metadata mark-up (Boiko, 2002, pp. 453-483), of all contents in the CMS database, specifying inter-relations among the contents. Combining metadata with the file structure of the CMS, the search procedure filters results and only returns a single link to a certain introductory file for each cluster of inter-related contents. Thereby, a typical search result is reduced to a few links; each one pointing to an introductory file that leads the user further on to inter-related contents in a structured manner. The method is more or less similar to that descibed by (J. Nielsen, 2000a, p. 231). This way in which to present and structure search results might also be usable for other websites similar to DBUR's in terms of emphasis on knowledge dissemination and a large proportion of inter-related contents. The presenting of search results is not to be confused with another feature of the reconstructed website, which is a subject index (Clarke, 2001). The subject index is constructed in collaboration between the earlier mentioned user-panel and DBUR's communication specialists, librarians and researchers.LimitationsThe reconstructed website only partly realises the potentials. First, it still lacks some of its originally intented features: A conference-like facility that makes it possible for users to comment on research results, share experiences etc. is missing, as is an online shop. Hopefully, these features will be realised within a few months. Secondly, it lacks some graphical fine-tuning: The fundamental concept of a simple, no-nonse graphic design is to be maintained, but right now it has become too 'bold'. Also, there are much too few illustrations on the website. Thus, a redesign proposal has been ordered with an external graphics design firm, and an illustration management system is on its way in order to make it easier for the editors of the website to use photos and other illustrations. Thirdly, the editing and metadata marking up of the content is not completed: Most visible when parts of the screen suddenly are empty. However, it is only a matter of short time until these things have been fixed. Fourth, the CMS behind the website is far from optimal: The CMS is unable to administer dynamic metadata, and therefore it is a tedious process to operate on the file structure. This is a severe problem as the search procedure heavily rests on metadata. Finally, while the above mentioned focus-group interviews identified a demand for computer-mediated access to information resources across the different knowledge institutions in the housing and construction sectors, the reconstruction of DBUR's website does not feature such a cross-institutional access. Parallel to the reconstruction project, a concept for how such a service can be establishing is described in the appendix to the thesis. This, much larger project, is under consideration by prospective financial partners for the moment being.Lessons learnedThe main lesson learned is positive: It can be of great value to involve the users in the reconstruction of a website, not only when it comes to validating and correcting concrete solutions, but also as an inspirational force in the creative process. While the first has been common in HCI-design for many years, the latter seems to be more rare. Thus, the project demonstrates a successful triangulation of different empirical methods. Another, less positive, lesson learned is that it is important in advance to specify in details the demands that a certain solution has to meet. Else, the delivered solution might end up like the one in this case: It works, but it is hard to maintain. This problem represents a dilemma, because the best solution often is invented through a process that facilitates feedback loops and trustful collaboration with those who are to deliver the technical implementation. The most common way out of this dilemma is to divide the process and start out with a specification phase, for example by consulting specification specialists. Another way out is to find a technical partner that you really can rely on, but obviously this is easier said than done. No matter what, it is an important lesson learned to address the dilemma more thoroughly than it was done in this case. Finally, a lesson learned is that one thing is ambitions, another thing is what is realistic to accomplish within the limited (although not necessarily scarce) resources at hand. For example it has not yet been possible to mark up all content transferred from the earlier existing website to the reconstructed website, as metadata and file structure are much more complicated to work with than what was intended and expected. This, however, is no new lesson learned.