Electronic information resources have become attractive additions to the information services offered in special libraries. Many of these services are increasingly being offered via the Internet. The creation of websites as tools for making sense of the glut of information that is available on the Internet is becoming a sensible activity, not for controlling this information, but simply for being successful in facilitating access to it. This paper is a case study in using the world wide web to create a sense making tool within the context of an art museum library.
Within the context of special libraries it has become increasingly important to offer electronic information resources to users of libraryservices. Libraries have facilitated access to electronic databases andelectronic resources such as CD-ROM (indexes, full-text, andmultimedia) for some time now. There has been a more recent trend toadditionally include Internet and World Wide Web (WWW) access tolibrary users. The reasons for adding Internet and WWW electronicresources to the materials offered by special libraries may range fromincreasing pressure from users to include value-added electronicservices, to political pressure within the organization to provideInternet accessible services in order to stay competitive, and in notfor profit institutions the impetus may derive from a gift or endowmentprompting the consideration of such services.
This paper endeavors to present the steps in creating a website fora museum's art research library. The process of planning, developing,and implementing the web site is described with the hope that it mayserve as a case study for individuals who are planning to developtheir own websites. While the author played a key role in the HTMLcoding, layout, and graphic design of this site, the criteria for theappearance and content of the site were based on the decisions of theparticipating librarians. During the development process, ideas,sketches, and finally prototypes were continuously submitted to participating librarians at key stages before proceeding. As such, thiscase study will only mention those stages which are salient to thefinal analysis of the site's development.
2. Description of the Website
The website for this art library was intended to be the gateway tothe Internet for users of the library's Electronic InformationResource Center (EIRC). The EIRC is a room within the library that has6 Windows NT workstations connected to a LAN. Internet access andCD-ROM titles, such as art CD-ROMs with full-text and multimediacontent (such as electronic exhibitions and catalogs) and referencedisks for indexing and abstracting services (such as the Bibliographyof the History of Art, ArtIndex, Mayers, etc.), are accessible throughthe LAN.
The EIRC workstations have web browsers configured to access a sitecreated by the Library. The content of the site consists of two mainsections: 1) General Information about the Library and the EIRC, and2) a catalog or index of Internet Resources. There is also a set of"Help" pages and a link to the library's online catalog. The contentwhich we are most concerned with in this paper pertains to the secondsection, Internet Resources.
The Internet resources section of the website was intended to be aservice which enabled art researchers to make sense of the content ofthe WWW by collecting Uniform Resource Locators (URLs) for sites whichwould be relevant to research interests pertaining to the visual arts.It is essentially a catalog of selected Internet resources created inHyper Text Markup Language (HTML) for the WWW.
The curatorial department delineation in the museum is essentiallya classification of art subject domains. It seemed logical, therefore,to classify websites using department names as subject headings. Thesite also cataloged general-reference categories of sites includingorganizations, other libraries and museums, electronic journals, sitescontaining ready reference materials, and visual resources (images ofart works in the visual arts) sites.
The project was proposed by the Chief Librarian, who appointed the Systems Librarian as the person to head design and development. Agraduate intern in the library at the time was identified as a goodparticipant for doing the HTML coding. The library would hire anElectronic Information Resources (EIR) Librarian in several monthsfollowing the initiation of the project. The EIR Librarian would takeover the development of content and user training activities after aninitial prototype for the site was agreed upon and after a fullytestable version could be released to users in the EIRC.
The audience of the website are a diverse community of usersconsisting of any Museum staff (from curators to guards), visitingscholars, graduate students, art dealers, auction house staff, artgallery staff, other museum staff, museum members, and anyone whoapproaches the library with an appropriate art research need.Additionally, the library intended to make the website available tothe public on the Internet, with the exception of restricting accessto subscription-based databases. It must be noted, however, that thelibrary considers the target audience to be the library users who are, generally speaking, researchers of art history whose information needsare at the graduate or professional level.
Filling information needs appropriate to the intended audience
The website selectively indexes art research materials on the WWWthat contain information which is as useful to users as theirnon-electronic analogs -- print information resources. Library usersexpect that information provided is culled from authoritative sources;this is probably the most important aspect of any new medium fordelivering information in a research/special library setting. In thecontext of the special library, it may not be necessary to have allthe innovations associated with multimedia in order to deliver valuable information. In the final analysis, users of the website willprobably hold that the most important criterion of any researchresource is to get correct and relevant information from the bestsources available.
The library hopes to increase user awareness of the many electronicinformation resources that are available through the Internet, such aselectronic journals and electronic reference materials, by pointing tothese resources from the website. While the uses of the Internet atpresent are changing at a rapid pace, it has become clear tolibrarians that the electronic networking of the world's informationresources through the Internet will make it possible to satisfy manyinformation needs more quickly through the Internet. It is hoped thatthis new electronic paradigm for document access (and documentdelivery) will be received as a user-friendly and robust innovationthat will broaden the materials available to the art researchers inthe library's user community.
Comparison to other sites currently available on the WWW
The librarians who initiated this project were well aware of other websites which were providing art research materials. Related sites atthat time included indexes and directories of art sites such as, arthistory instructor Brian Witcombe's "Art History Resources on the Web", the "Gateway to Art History", which is patterned after Gardner's Art through the ages, the University of Kentucky's ArtSource Page, and the Museum of Modern Art's page of resources in the arts. It was hoped that the library's site would incorporate some of the characteristics of these popular indexes, but would do so in a manner reflecting the scholarly needs of the museum staff. Internet resourcesthat were indexed would therefore need to reflect the research scopeof the museum that is delineated by the curatorial departments.Particular attention was paid to the selectiveness of catalogingInternet resources. It was also very important that the appearance ofthe site reflected the corporate image of the museum, that theinterface was user friendly, and that the site's contents were easy touse and make sense of -- a function of information organization. Particular attention was placed on making the organization of thewebsite reflect well-understood library information structures such ascatalogs and indexes.
Multimedia elements to enhance the resource
Members of the development team were aware of the possibilities forusing multimedia elements to enhance the site. The use of multimediafor conveying art research information is particularly valuable inresources which make use of images. Exceptional examples of the use ofmultimedia include visual resources databases and sites which containQuickTime VR documents showing objects and architectural spaces inthree dimensions. It was clearly established, however, that thelibrary's site should only use these elements when necessary for thecommunication of information and to represent navigational actionsthrough the interface. Multimedia aspects such as sounds and videowere to be avoided if necessary. It was agreed that these elements would mainly embellish the site, and would not be necessary for a sitewhich serves predominantly as an index or pointing tool for findingother sites. Text, therefore, is the predominant format in the site.Images are used to convey hypertext functions of the site, and tocontrol the appearance of the menus due to the variable treatment offonts in different WWW browsers. Images are additionally used on thehome page to show the museum's logo (a renaissance "M" within acircle) and to show one of the museum's art objects on the "Internetresources: Art-subjects page".
3. Description of the development process
The following section is a discussion of the development process.It is concerned with how the work was accomplished and how decisionswere made. It is hoped that this section will provide the most usefulinformation to special libraries. The enumeration of steps undertakento develop this project and the observations of what went right andwrong may be valuable to librarians who hope to avoid making similarmistakes in their development cycle.
How the work was accomplished
The work of planning a website was initiated in a meeting involvingthe Chief Librarian, the Systems Librarian, and the Intern. Followingthis meeting, the purpose of the site and the market for the site wasestablished for the Website development team, which consisted of theSystems Librarian and the Intern. It was decided that the site wouldcontain information about the library and the EIRC, but the main drawto the site would be a bibliography of art research resources on theInternet. Several brainstorming sessions were held by the SystemsLibrarian and Intern. During these sessions, ideas were shared todefine the information the site will contain, what the site should look like, and how one will navigate through the site's contents. TheSystems Librarian began to raise questions and suggest issues relatedto the following: 1) structural design, 2) visual design, 3) technicalissues related to creating the content, and 4) issues related to theselection of the content and the recording of bibliographic data.
Exploration of these issues helped give form to elements of thesite. These issues will be discussed below. In order to make sense ofthe development process, it should be noted that following thebrainstorming sessions, ideas were executed in HTML by creating usable"prototype" web pages. It was useful to follow a prototyping model ofdesign development because participants could see the site as it tookshape and comment on the elements which they approved or disapprovedof. Participant feedback was highly encouraged and figured greatlyinto the design of the site.
The structural design of the site was to meet specificationsidentified in the brainstorming sessions. The site should avoid havingtoo many levels of navigation. Therefore, the hierarchy of sectionsand subsections should be limited. To avoid the difficulty ofnavigating within a hierarchy, a menu should be available on everypage of the site, in order to flatten the hierarchy and permithypertext linking to any of the main sections. The main sections arethe access points that were accessible from the home page. This menuis placed in the upper left portion of the screen, which we referred toas the "gutter" -- a term appropriated from the word used to describethe portion of page in a book towards the inside seam or spine.
The final structure of the database consists of very few levels ofnavigation to arrive at any point in the site. At the highest level(the home page), one can only access the six main categories of thesite: 1) General Information on the Library or 2) General Informationon the EIRC, 3) Internet resources: Arranged by art-related subjectsor 3) Internet resources: Arranged by general subjects, 4) Watsonline,5) Help, or 6) Feedback. (See Figure1 to view the visual display of thehome page) The descriptions of these categories are somewhatself-explanatory, so the team believed it would not be necessary toexplicitly describe the contents of each section. For instance, theheading "General Information" lists only the Library or the EIRC. We thought that these sections would not need further explanation. Theheading "Internet Resources" lists only "Arranged by art-relatedsubjects" and "Arranged by general subjects". The description of thearrangement of the resources gives the user a sense of the contents ofthese pages and is a cue to help them make a decision regarding whichdirection to take at the home page. The Watsonline link is explanatorymainly to users of the library. It would probably have been better toinclude a parenthetical phrase to read, "Watsonline (online catalogfor the library collection)", for users outside of the physicallibrary who will want to use the catalog may not know the meaning ofthe term "Watsonline". The help feature has become somewhat commonplacein computer applications and is self-explanatory. In the menu, linksto all of these sections are graphically rendered as text, rather thanas icons, in order to avoid confusion. A book icon is used torepresent the home page. The book appears with the word "home"rendered in blue beneath it. A blue "M" button appears at the bottomof the menu to return the user to the museum's website.
Figure 1, Home page
Navigation in the site requires taking several steps (makingseveral clicks) through the hierarchical organization of the site'scontents. The "steps" represent the organizational "chunks" or"clusters" created by the development team, so that users could makesense of the material contained in the site. Each step gives the useradded information about the contents of the site. For instance, whenone clicks on "Internet resources: Arranged by art-related subjects",a table of art-related subjects is displayed. (See Figure 2) Selectingone of these subjects, such as "American Art", will display a table ofresources related to American Art. (See Figure 3) From this table,one may click on a URL to link directly to the Internet resource, orone may click on the magnifying glass icon to get a better descriptionof the site. Clicking on the magnifying glass link for "AmericanMemory from the Library of Congress", for instance, brings up a descriptive index record for that Internet resource. This recordcontains the following information: Author, Title, Notes, Description(or abstract), Language, Category, and Keywords. (See Figure 4).
Figure 2, Table of art subjects
Figure 3, American art subjects
Figure 4, Site record
An important step to making sense of where all of these sections would fit into the scheme of pages in the website was to create a storyboard or flow-chart. This diagram helped us map out where information should reside; it helped us flesh out the structure of the site. The diagram also serves as an illustration of thedecision-making process in using the system. It allows us to visualize the user's cognitive movement through the website. One of the finaldiagrams is visually illustrated in the screen below. The splashscreen at the top was eliminated.
The team believed that the goal of keeping navigation simple wasaccomplished. The greatest number of clicks to get to any one sectionin the site is four. For example, to reach the "American Memory..."bibliographic record above one had to click four times. It ispossible, however, to get to the site in three clicks if onerecognized the title in the resource table. It is hoped that the sitewill additionally offer searching of the bibliographic records by fieldin the future so that users who know exactly what they want to searchfor may retrieve pages without navigating through our pre-coordinatehierarchy.
Before planning the visual appearance of the site, the team lookedat other sites which were doing similar projects and looked at sitesdiscussing user interface issues. A very influential resource whichwe considered was the Sun MicroSystems website because of its treatment of user interface and human computer interaction (HCI)/usability issues. Print resources including two texts from Edward R. Tufte's series (1990, 1997) on the visualization of information were also quite valuable. Several important criteria emerged from our research. The site should integrate with the image and color scheme portrayed in the museum website. It should avoid using striking colors, electing instead to using color combinations which are legible on screen and which complement each other. The content of pages should generally appear within one screen set at a resolution of 600 x 480 pixels (the lowest common monitor resolution for 14 inch monitors excluding laptop monitors). Designing for the screen at this resolution ensures that users will not need to scroll to access important information. It is well understood in the HCI literature that many users generally do not like to scroll or simply do not know that there is content below what is visible on screen. Finally, in consideration of users accessing this site with slow modem connections, the use of graphics was somewhat restricted. Graphics, it was decided, should be few, and should be utilized only in significant areas of the site. Where they are used they should be small to ensure quick downloading.
It was apparent that the purpose of the site was to beinformational. The site was to convey content as efficiently andeffectively as possible. As a result, the site has a markedminimalist and corporate appearance. The use of graphics is sparse.The areas where graphics were used were mainly on the home page and inthe menus. The home page could have been rendered completely in text, however, it was implied that the site should look somewhat impressive, reflecting in appearance the arrival of a new innovation for thelibrary. The home page and the menus were rendered almost completelyin graphics using tables to ensure that the appearance was consistentregardless of how it was being accessed, or regardless of what fontswere being used to render the page in a browser -- using graphicseliminated the inconsistency that comes with variable font sizes inbrowsers. It must be noted, however, that the earliest browsers (suchas Mosaic) render pages somewhat differently because of the use oftables, but, the information on the pages remain intact.
Technical issues and the bibliographic tool
One issue that was somewhat problematic when doing the projectinvolved the creation and maintenance of the bibliography of Internetresources. The team wanted to plan for the growth of the bibliography.It wanted to ensure that web pages were stored in a sensible directorystructure that reflected the organization of the site. It also wantedto consider the possibility of somehow smoothly exporting all of thebibliographic records into a database management system (DBMS).
In the first several prototypes of the site, the main access pointswere developed: the home page, the art and general subject pages, andfinally the resource tables within each subject. We initially createdresource tables for each subcategory under a subject by collectinglinks, creating a brief description for each site, and placing them ina list rendered in HTML. However, we discovered that the project wouldbecome unwieldy if we decided to include sites under severalcategories. A page such as the "American Memory from the Library ofCongress" can be placed in "American Art" as well as in "VisualResources", for instance. Initially, we believed we could create cross-references in the form of "see" and "see also" references, or wecould simply duplicate the listing in the other tables. This wouldcreate some difficulty in time if we were to catalog many sites.
The Systems Librarian was aware of a bibliographic applicationowned by the library that allowed importing MARC records into a simpledatabase for reporting purposes. Records could be created using anumber of workforms or templates set up for different bibliographicformats. Sets of records may be searched using Boolean operators andselected separately. These records could then be printedalphabetically in a number of pre-defined styles including APA, MLA,and Chicago, or custom styles could be generated by making a few modifications to the output style. The application is ProCite forWindows, which runs on Windows 3.1, 95, and NT operating systems.
ProCite turned out to be quite robust, and functioned essentiallyas a flat-file database. It allows the capturing of a website beingaccessed by a web browser running in the background. Upon selectingthe "capture page" feature, ProCite creates a new bibliographicrecord, shows the ASCII text contents of the captured page in onescreen for copying and pasting into fields, and pastes the URL of thesite into a "storage/location" field. It was discovered that theoutput styles could be generated into ASCII or HTML text documentswhich could be saved to disk. This gave the team ideas for automating the HTML work they were doing in the prototypes, and thus saving timein developing the website. As it turns out, the application not onlyhelped facilitate creating bibliographies in an automated fashion, butalso allowed the team to use controlled vocabulary terms in theircataloging, and allowed the team to generate complete web pages usingthe application.
The ability to export HTML files was quite a revelation. But theoutput styles generated using the pre-defined formats were quiteunusable. The records would have to be embellished with theappropriate menu links, and the names of the fields that were outputwould need to be renamed. The Intern began wondering how functionalthis feature would be given the the amount of cutting and pasting ofHTML that would be required. The team explored ways of custom coding output styles, and came up with a solution for exporting individualrecords in HTML with all of the appropriate menu link graphics andfield names. Exporting required highlighting each record, exportingthe record as an ASCII file, renaming the extension ".htm", and savingto a directory of records. It would have been quite valuable if wecould create a macro that did all of the exporting and named each fileaccording to the record number field we established. But this task wasnever seriously pursued. Additionally, creating the resource table wasa process that required more steps, and was a little more complex thanthe exporting of records. Exporting tables required searching for acategory in the ProCite database, such as "American Art", exporting a custom table into a text file, and finally copying and pasting thattable into a template for the category. This multi-step process wouldadditionally need to be repeated as records were added to the subjectcategory.
The amount of time saved in outputting the records into HTML fromProCite instead of hand coding each record is inestimable -- there arecurrently over 220 bibliographic records alone. However, thepseudo-automation solution created for this website using abibliographic tool was seen as an interim step towards creating afully functional DBMS that could be accessible from the website. Itmust be stressed to staff at the library that this is the intention,so that the initial goal of creating a DBMS is not forgotten. Should the bibliographic database become too large and unwieldy for librariansto maintain using the current tool, a move should be made to migratethis data into a more robust information management system.
4. Recommendations for further development
The introduction and training sessions given by the EIR Librarianhave ensured that the website will be received positively by the usercommunity. Suggestions to catalog new sites are being welcomed fromusers. The overwhelmingly positive reception of the website as anadded service for users has been quite rewarding. However, it will bealso valuable to view the project in retrospect, and to comment on theaspects of the project that did not work, as well as citing theaspects that did. At this stage of use of the website, criticism hasbeen minimal, but feedback from users in the near future will help inpreparing for future releases and revisions.
What was effective/productive and ineffective/non-productive
Developing the website by planning and prototyping was quiteeffective for this project. It cannot be stressed strongly enough howmuch the planning phase has impacted on the development of thisproject. Planning the website consisted of many question and answersessions between the website development team and the various otherparticipants. It would have been extremely valuable to have spent moretime, however, in creating a complete document to define the project.Additionally, it would have been valuable to set more clearly defined goals and deadlines for stages of completion, and to schedule regular assessment sessions with participants at key stages.
An aspect of the development process that may have been inefficientwas prototyping the initial pages completely in HTML. The first fewprototype web pages were quite easy to put together; a few versions ofthe home page had been completed and were approved for posting.However, a political issue required that one of the headings on thehome page be changed from three words to twelve! The adding of ninewords completely changed the appearance of the page, and members ofthe staff were additionally suggesting that names of categories bechanged. The entire home page had to be redesigned. Given that the 12word heading could create problems because of fonts, the Interndecided to render all of the home page elements completely in graphicsto ensure complete control over the font size and layout.
After some time, it became apparent that it would be best to sketchout prototype pages on paper before creating entire pages in graphicsand HTML that may need to be scrapped. Given that graphic designtools were limited and that the deadline for rolling out the websitewas quickly approaching, sketching on paper became the developmentparadigm for the home page layout as the work of filling outbibliographic records progressed. In the end, the final home page thatappeared on the site was one of two sketches created over lunch on acocktail napkin by the Intern.
What remains to be done
Migration to a more robust information management system, such as aDBMS connected to the web server will be the final step to completethe project. This step will prepare the website for a second, improvedrelease. It is unlikely, however that this will happen, given theamount of control the museum's systems department places on securityissues. The work of filling out bibliographic records progressesdaily, and these records must be added to the server on a periodicalbasis.
Features that were not implemented because of time constraints
Preparing this paper after designing this website was a very valuable exercise for me. As a case study, I believe it serves to illustrate the value in taking measured steps towards developing a multimedia application such as a website. Shortcomings that may be detected in the design of the site are largely due to oversights in the planning process. Tasks that may have become difficult to dealwith if not considered, including simple issues such as the organization of the file structure and the naming of files, were handled tactfully because they were planned in advance.
It is important to define the purpose of the site intellectually before acting on a plan. The development of a web site is much like the creation of a software application. The goals for using the application must be conceived ofand defined, a strategy must be created, and only then should one begin enumerating the tasks that are essential to designing the appliaction. With the mission, goals, purpose and strategy documented, one may proceed to create a successful application.