Museums and the Web 2005
Papers
The modular componants

Reports and analyses from around the world are presented at MW2005.

The Virtual Lightbox for Museums and Archives: A Portlet Solution for Structured Data Reuse Across Distributed Visual Resources

Brian Fuchs, Max Planck Institute for the History of Science, Berlin, Germany, Leif Isaksen and Amy Smith, Ure Museum of Greek Archaeology, University of Reading, UK

http://www.rdg.ac.uk/ure/VLMA/

Abstract

Virtual Lightbox for Museums and Archives (VLMA) is an RDF-driven visual collections aggregator/syndicator applet that allows viewing, collecting, and reusing distributed visual archives and relevant metadata via P2P technology. It is a response to specific practical problems in content integration and reuse encountered in digitizing and publishing a small museum's collections and in adding them to larger portals. In small collections object types are normally represented by single examples, if at all, so that such institutions are crucially dependent, for both teaching and research, on comparisons of their holdings to those in other museums and archives. On-line resources could provide much of the requisite comparanda, yet differences in presentation from Web site to Web site severely limit this potential, as does the well-known difficulty of maintaining references to off-site data. To address this problem, the VLMA has developed a portlet approach, in which collections with intrinsically heterogeneous metadata sets are syndicated and their contents collected – browsed, stored, viewed, and reused – at the peer/client level on an object-by-object basis. This allows metadata integration to be performed at the point of reuse, by the end user, an approach which complements more traditional ones such as common metadata structure (CDWA) or metadata aggregates (OAI). Content reuse can take several forms, ranging from a presentation to resyndication of collected objects in the form of a new collection. The latter possibility provides an easy method for bringing added value to published content as well as a simple way of creating thematically related collections with distributed content.

Keywords: syndication, P2P, RDF, lightbox, metadata, applet, federated searching, interface

Introduction

The Virtual Lightbox for Museums and Archives (VLMA), which is being developed by the Ure Museum of Greek Archaeology (University of Reading) in collaboration with the Max Planck Institute for the History of Science, Berlin (MPIWG), is an RDF-driven visual collections aggregator/syndicator applet that allows viewing, collecting, and reusing distributed visual archives and relevant metadata via P2P technology. It is envisioned as a low-cost method of publishing electronic databases – particularly those that are rich in visual content – on the Web in a manner that will make them easily navigable and useful to researchers, teachers, and students alike. It has evolved in response to specific practical problems in content integration and reuse that the Ure Museum staff encountered in digitizing and publishing its collections and in adding them to the ECHO humanities portal (http://echo.mpiwg-berlin.mpg.de). In the first half of this paper, after an introduction to the Ure Museum and its database, we will investigate the reasons that museums go on the Web, and the challenges they face in making their materials available on the Web; then analyse whether or not previous solutions have satisfactorily addressed their needs or answered the inherent challenges. In the second half we will introduce our portlet solution, the VLMA, by explaining its functionality, structure, and means of implementation.

The Ure Museum: A Case Study

The Ure Museum of Greek Archaeology, founded ca. 1922, has survived in relative seclusion for a half century in its current home within the Classics Department at the University of Reading; its digital history is less than ten years old. A recent effort to open up its resources and facilities to the University as well as the wider community is afoot. It follows on the heels of the Ure's achievement of registered museum status in 2001, under the Re:source scheme (http://www.mla.gov.uk/documents/musreg_eng.pdf), and its Widening Participation case study, in 2003, sponsored by South East Museum, Library and Archive Council (http://www.semlac.org.uk/casestudies_urefull.html). The focus of this audience and access development that is relevant to the origins of the VLMA is the Web publication of a database that opens up the museum contents to all potential visitors. Virtual access to the Museum should help students – whether at Key Stage 2 (aged 8-12), AS- or A-level (aged 16-18), University (aged 18+), or beyond, to prepare for and follow up on their visits to the Museum or to prepare projects based on the collection (or parts of it). More elaborate applications by teachers, researchers, and scholars are also envisioned.

The digital history of the Ure Museum began in 1998, when Amanda Hart, then Assistant Curator, created a Ure Museum database (hereafter Ure DB) in Access. Amanda hardly found the time to input data. Since 2000, student involvement and volunteerism at the Museum has enabled the development and maintenance of the project. A pioneering undergraduate, Peter Cunningham, approached the Museum in 2000 with an enthusiasm for databases, and was soon put to work on the onerous task of typing ca. 1400 entries from the handwritten (and virtually illegible) card catalogue into the Ure DB. Collaboration with Brian Fuchs at the Max Planck Institute of the History of Science began in 2001, when he worked with the Ure Curator, Amy Smith, to create a platform independent “bespoke” database structure. This database, written in pure Perl, allows multiple simultaneous users to access the Ure DB on the Web: staff, students, and others may freely consult the data while select users (with password access) may edit it. The related data sets that underlie the Ure DB provide access to digitised images of artifacts as well as archival materials, most of which have been obtained at no cost or with limited incentive income through the efforts of students and other volunteers.

As the database neared completion in 2004, we began to research and plan the implementation of user interfaces and thus revisited the question of why museums go on-line, besides publicity and internal benefits. We encountered the usual problems of second-guessing the users: Who were they? How would they want to view and use the materials? What would they use them for? Feedback obtained through our Widening Participation project (see above: http://www.semlac.org.uk/casestudies_urefull.html) suggested our initial inclinations, towards a variety of user interfaces designed for specific age groups, were wrong. Users at all ages and academic levels longed for (a) the greatest flexibility, to go beyond the tombstone information provided about objects in traditional museum labels, and dig as deep as they wished into the database; (b) images that provided multiple views as well as details of artifacts. Users craved such access both to be able to find and study, on the one hand, one object in detail, and to have sufficient information to be able to compare, on the other hand, two or more objects.

The inherent interest in comparing museum artifacts is perhaps the best reason for all museums, especially those with limited collections, to publish their collections on the Web: research categories cross collection boundaries. Two examples of this phenomenon that are particularly relevant to the Ure Museum are the need to (a) (re)combine distributed objects or (b) combine distributed assemblages of objects. Examples of ancient art, like all archaeological materials, are fragments of the cultures that they represent, and usually fragments of the objects from which they come (Smith, 2000). Quite frequently a scholar might find a fragment of a sculpture or vase in one museum that joins to a similar piece in another museum. Dyfri Williams has done just that with an Archaic Greek vase fragment, in the Ure Museum (inv. 26.2.1), that joins a dinos (bowl) attributed to the painter, Sophilos, which is housed in the British Museum (inv B601.26 or B100: Williams, 1983). So access to the fragment on the Ure DB gives visitors only a glimpse of the whole, and to see the more significant parts of the vase, one has to have access to the corresponding piece in the British Museum (it is, however, absent from their on-line Compass provision: http://www.thebritishmuseum.ac.uk/compass/).

The same Archaic fragment is also part of several distributed assemblages of objects. For example, someone interested in the works of Sophilos would wish to consult all of the 91 works attributed to or signed by that artist: the single fragment on the Ure DB, and the more impressive objects in the British Museum, Greece's National Museums, and elsewhere. These are fortunately brought together, albeit in limited form, on the Beazley Archive (http://www.beazley.ox.ac.uk/), which for reasons of copyright only reaches a restricted (scholarly) audience that has applied for and been granted password access. One might also be interested in studying the other vases and vase fragments that, like our Sophilos dinos, were found at the great Archaic Greek trading post, at Naucratis, in Egypt. The British Museum is now going through the painstaking task of gathering this assemblage, virtually. Meanwhile most of the relevant material is not available on-line, and barely published in any format. With these examples we hope to have shown that a small collection such as the Ure Museum, where an object-type is normally represented by a single example, if at all, is crucially dependent, for both teaching and research, on comparisons of its holdings to those in other museums and archives.

Challenges in Museum Web Provision

Textual And Visual Comparisons

On-line resources could provide much of the requisite comparanda, yet in today's climate, with each major museum rushing to create its own 'docu-island' and justifying expenditure on visually rich but textually uninformative Web sites, the Web enforces rather than breaks down the boundaries across which research categories extend. Quite simply put, just as it is impossible to answer research needs within one museum, it is impossible to answer research needs within one museum Web site. Differences in presentation from Web site to Web site severely limit this potential for researchers, let alone lay audiences, to cross institutional boundaries. There is also the familiar difficulty of maintaining references to off-site data. Some work on the Semantic Web seeks to accommodate heterogeneous database content (E. Hyvönen et al., 2004; for more general studies on the Semantic Web, see Berners-Lee, 2001, and Fensel, 2002) although the visual element is often lost in such discussions.

Useful comparison of museum materials relies on the opportunity to compare images as well as other information (usually textual) about the artifacts that they represent. Image comparison technology is widely available to professionals who prioritize images in their work – architects and designers, other artists and technicians, to name a few: the Virtual Lightbox (http://mith2.umd.edu/products/lightbox/applet.html), an open source project developed by the Maryland Institute for Technology in the Humanities (MITH: http://www.mith.umd.edu/) provides users with an opportunity to collect various images and then to sort and/or modify them in relation to each other, as on a traditional (non-digital) lightbox or light table. VLMA adapts this useful open source applet to the museum and archive sector by enabling the user to retain metadata (text concerning the image as well as the object represented by the image) while viewing and sorting the images. The retention of metadata is crucial to the applicability of lightbox technology to museums, for the purposes of teaching and learning, despite the fact that museum Web sites are as guilty as any of presenting images with little or no textual reference to the artifacts that they represent, or in a manner in which viewers could inadvertently lose (close) the textual reference.

Web Site Choreography

There are costs to on-line presentation which remain hidden to the producers of Web sites because they only become visible among the consumers of Web sites, and even then only when consumers compare notes. A good example is an area we can call 'Web site choreography': the whole process of designing and implementing a site, as well as (from the consumers point of view) of deciphering, learning, and turning into 'second nature' the complex series of steps and gestures involved in fluent Web site use and navigation. On the producer side, a great deal of time and money has to be spent designing a Web site. Dedicated staff have to be hired for this purpose; and even when the project is out-sourced, higher level managerial involvement is certain to remain lively, as the Web site represents the 'public face' of a collection or institution.

To be sure, the intention of designers is to make visiting a collection's Web site a pleasant and rewarding experience; but for the consumer faced with an ever-growing number of such highly choreographed on-line resources, the situation can seem more like anarchy. The more complex the decisions taken in implementing a Web site, i.e. the more staff time implementation consumes, and the further the decision-making process traverses an institution's hierarchy, the more 'different' a Web site is likely to be from other Web sites with similar content. As a result, users each have to construct a new 'mental model' of the choreography for each site they visit. And if a site changes its choreography, they have no choice but to throw out the old model and construct a new one. Nor are mental models, once constructed, easily transferable to other sites: users can perhaps teach others how to use a Web site effectively via 'show-and-tell', but programmatic, machine-readable communication of what has been learned is typically out of the question.

Nor is that by any means the end of the story. Another hidden downside of intensive investment in Web site choreography – and one that is perhaps more disturbing from the point of view of data providers, however often it may be ignored in reality – is potential bias in data presentation. Elaborate thematic groupings, for instance, which are useful for highlighting connections between collection objects, also obscure other possible groupings from view and use by making the chosen groups ergonomically preferable. A user accustomed to working easily with a collection of objects thematically grouped under, for example, Greece: classical period will also be inclined to favor a mental model reinforced by access habits. The result is a production cycle that creates something which is hard to learn, expensive to produce, inconsistent across Web sites, transient, hard to abstract, difficult to reuse, and potentially biassed as to the underlying data.

Content Negotiation

But even when Web site fluency is not taken into consideration, content negotiation remains a problem. It often remains less than certain that users will be able to extract from a collection Web site the information in the required form. Immediately they are faced with questions such as “Can I download the data I need?” or “Are the images for collection objects easily found / accessible?” Even if users have managed to extract the desired information about objects from a collection – perhaps they have found the image and discovered its relevant metadata (that is stored elsewhere) – it is by no means certain that they will be able to work with the material collected. Often the metadata is essayistic, and therefore inconsistent from one item to another. And if metadata about an object and image data are stored separately on a Web site – which, for technical reasons, is the norm – each user must negotiate a personal local storage solution – a situation which normally leads to storage in ad hoc, task-specific, and ultimately non-reusable form.

Previous Solutions

Thematic Portals

To overcome these problems, data providers have often turned to thematic Portals – attempts to collect data on a particular theme that is scattered across cyberspace and to re-present it in a consistent fashion, in such a way that objects that belong together thematically, or represent disparate assemblages as discussed above, actually appear together (Crane et al., 2001). There is no denying the usefulness of such 'consolidation' tools, which allow users to visualize requests from several different and heterogeneous sources on a single Web page. One of the most common, and most successful, examples is the library portal, which allows users to search through the Web catalogues of the world's libraries with a single search box (e.g. http://www.ubka.uni-karlsruhe.de/hylib/virtueller_katalog.html). In this case, however, users leave the portal with the search results, and are again faced with the task of figuring out how each individual catalogue site wishes to be used. (On the general problem of inconsistency between portlets [portal components] cf. Priebe and Pernul 2003, which, however, reflects the general tendency to see the problem from the standpoint of the data manager – inconsistent data – rather than that of the end-user – the reusability of inconsistent data and of skills learned in coping with different data presentations.) In order to cope with the problem of heterogeneity, a portal may attempt to incorporate material from disparate sources directly into its Web site and perform certain analytic operations there: both the Perseus Project (http://www.perseus.tufts.edu) and European Cultural Heritage Online (ECHO) attempt this (http://echo.mpiwg-berlin.mpg.de). While this approach allows the portal to generate integrated metadata and connectivity, it inevitably does so at the price of the annexed project's ability to present its metadata in a form that it considers fully adequate to the demands of content.

Common Metadata Standards

Where portals threaten to 'homogenize' presentation, common metadata standards have seemed to offer a less intrusive solution. By following a common metadata standard, a collection can 'buy in' to a common pool of information and therefore of access potential, and allow users to find and bring together widely distributed objects that have thematic similarity. Yet the fact is that domain-specific common standards have not been widely adopted: there have always been more preachers than converts. This surely results from the fact that they are costly to implement and are presented to museum data providers in an intimidatingly complex manner. There is an obvious trade-off: the finer-grained the standard, the more complex it is to implement; but the easier it is to implement, the more constrained a collection's metadata become. Moreover, domain-specific metadata standards do not even recognize, let alone address, the problems posed by Web site choreography as noted above.

Distributed metadata solutions (such as the Open Archive Initiative http://www.openarchives.org/) seem to offer a way out of this impasse by allowing collections to provide their own metadata schemas, alongside a 'standard' mandated one, such as Dublin Core. It thus becomes possible to maintain metadata integrity while reaping the advantages of a common, albeit general, metadata format. This approach represents a genuine advance, as it makes possible, for the first time, standards-driven federated searching over large heterogeneous corpora. The drawback is, however, that federated data-harvesting, however attractive it may sound, is not a function to which the consumer has any direct access. The data harvesting is implemented and therefore necessarily repackaged by participating data providers; this again brings into play the entire problematic of Web site choreography.

A similar limitation inhibits the usefulness of the meta-metadata technology – ontology mapping and inference technology such as OWL (http://www.w3.org/TR/owl-features/) or domain specific CRMs (Conceptual Reference Models) such as the CIDOC CRM – which can be used to enhance and focus the distributive metadata solution and are currently the focus of intensive development. By using a CRM such as CIDOC-CRM (http://cidoc.ics.forth.gr/index.html), which is developed for the museum community, a collection can tap a common semantic reference point for recurring categories in collection metadata within a large domain. But again a great deal of time must be invested in refitting collection metadata to take advantage of such advances; and even when the investment has been made, the decisions about how to present metadata remain in the hands of the data provider.

The VLMA Solution

VLMA provides an alternative solution, one which is intended to complement rather than to compete with existing portal solutions like ECHO or metadata initiatives such as OAI (http://www.openarchives.org/) and CIDOC-CRM (We call this solution a portlet solution. By portlet we mean a resource whose function is to provide access to distributed information systems at the level of the consumer of information rather than, as in the case of other solutions, at the level of the provider. Normally a portlet signifies a portal component that sits on a server and provides a presentation layer to portals by means of a generic API (e.g. in Java enterprise systems, JSR 168: http://www.jcp.org/en/jsr/detail?id=168). Portlets constructed in this manner are highly portable, and can be used in a plug and play fashion with any conformant portal, but they still do not escape the problems with Website choreography and data inconsistency that we have described above.

VLMA takes the portlet metaphor a step further by freeing the portlets from the server environment and turning them into free-floating user-agents. The VLMA behaves as an independent consumer-peer that resides in a network made up of consumers and providers. Like a portal, it is capable of accessing a diverse set of collections; unlike a traditional portlet, it is not a component of any one portal, but rather an aggregation point for collection data that is located in the area of the information consumer, rather than the information provider.

This approach differs from most of the aforementioned attempts to integrate content in that it leaves to the consumer the task of negotiating metadata consistency. In this strategy, the data consumer, rather than the data provider, decides how to reuse data. The VLMA approach is to take a collection consisting normally of objects that have associated metadata and visual resources and to 'syndicate' them – that is, describe where to find them and what tools to use to access them – using RDF (Resource Description Framework: http://www.w3c.org/RDF/). The RDF consists entirely of aggregation points for material – visual and metadata resources, as well as tools such as RDF parsers and datastores already available on the Web (Lasilla and Swick, 1999). (A tool, currently available in prototype form, guarantees a low entry threshold by allowing data providers to generate and publish the required RDF by inputting a simple tab-delimited file that identifies where image and metadata resources are.)

The result of this syndication is two-fold. First, to the consumer, all collections look the same; they are simply bags of collection objects that can be collected for local reuse, without intervention from distracting, costly, potentially biased html presentation. Secondly, by syndicating services rather than contents of collections, we avoid a presentation based on ontology. The VLMA knows only about accessor services, and is blind with respect to the material being accessed.

What the user sees in the applet is a network of collections, each of which exposes its collection as a group of access services. The three basic services are Browse, Search, and Friends. The Browse service allows browsing of the entire contents of a collection, while the Search service enables structured searching through a collection's metadata. What appears within both services as the central unit of transaction is the Collection Object, which consists of the cluster of images and metadata fragments attaching to an object. This can then be 'collected' by the user into his own collection for further use.

One of the collections visible to the user is his own ('MyCollection'), which offers the same services as the collections of data providers, with one exception. In the place of 'friends', there is a lightbox service which contains objects that have been collected to the lightbox panel and may be viewed and compared there. The federating function of the Friends service is taken over by the Search service, which differs from the Search service in non-local collections in allowing the user to do a federated search over all of the collections that the user has discovered on the net by using the Friends service. It is also possible to annotate objects. A Web service permits collected objects to be exported to OpenOffice presentation format (Impress).

Functionality

The process of federation described above means that a user is able to navigate not only through the collection hosting the VLMA, but also through 'friend collections', which in turn offer access to other friends, and so on. This provides three key advantages. First, it makes the users aware of correlative collections of whose existence they may be oblivious. Secondly, it provides far greater ease of navigation: there is a homogeneous user interface across collections, regardless of the heterogeneity of the data. Finally, it allows for federated searching. Although search criteria remain limited in the current phase of development, it is possible to use VLMA to search for key terms over several different collections simultaneously and to return the results in a single set.

The next crucial component is the provision of tools for the examination and comparison of retrieved data, whether textual or visual. A lightbox panel enables the users to arrange, resize and manipulate their own selections of images. As the lightbox naturally does not edit the images itself, they can also be restored to their original forms at any time. Text comparison tools are more basic but allow for the hiding of empty attributes. As the metadata is stored internally in XML, support for XSL/XSLT (powerful text manipulation and presentation technology: http://www.w3.org/Style/XSL/) is foreseen in future development.

Finally, it is important to be able to extract selected data and store it in several ways: as state, in machine-readable form, and in presentation format. The lightbox does the first two by exporting to XML, the W3C-recommended format for data storage and transfer (http://www.w3.org/XML/). The file can be saved to the users' systems, enabling them to recreate their own personal collections in future, regardless of which host site is being visited. The format is formally closed and text-based, making it easily readable by other programs, as well as by human interpreters. In providing for presentations, we have enabled export to the presentation component of Open Office, an open source, multi-platform, office productivity suite developed by Sun Microsystems (http://www.openoffice.org/). (Those currently wishing to export to MS PowerPoint can do so simply by exporting from Open Office.) Although we are aware that its use is not as widespread as some other proprietary systems (e.g. MS Office), we felt it was crucial to make it possible for all users to access the functionality, even if that requires an extra (costless) download on their part. There is no reason why future versions should not provide support for a variety of other systems.

Structure

The VLMA is a multi-tier application distributed over the Web. It also contains elements of peer-to-peer technology which lead us to prefer the terms 'provider' and 'consumer', rather than 'server' and 'client'.

Schema of the constituent parts

 Fig 1: Schema of the constituent parts of VLMA and the functionality they provide

The provider is not in fact a single running process, but rather a series of 'responsibilities' that an agent making a collection accessible to the VLMA must undertake. A suite of tools enabling the provider to do so is provided by VLMA. The first task is to syndicate the provider's collection in Resource Description Framework (RDF) format (http://www.w3.org/RDF/). This gives everything denoted as an 'item' a Unique Resource Identifier (URI) (http://www.w3.org/Addressing/). This precaution is taken to ensure that henceforth all references to the item will apply to a single location and enables the collection provider to maintain control over the metadata corresponding to it. VLMA provides a scripting tool which, after initial configuration (usually a couple of hours) can be run automatically to update the RDF when changes are made to the originating database. The files are served by a Sesame database, the location of which is specified in a top-level RDF file (http://aduna.biz/products/sesame/index.html).

The next task is to provide services, by generating an RDF description for each service. Services currently take the form of Browse, Search and Serverlist. RDF provides information to the consumer on how to access these services. Browse gives the user direct access to all the items in the collection. As this could amount to many thousands of items, the information is 'chunked' so that only a subset is read from the provider at any one time. The browser can sort by any metadata field given in the collection. The Search function provides more rapid access to a specific subset of the collection. A simple search mask enables the user to generate a search condition containing multiple terms and addressing several fields. Finally, Serverlist informs consumers of other 'friend collections' and gives their location so that they too can be browsed in VLMA.

The last responsibility of the provider is to give the location of a variety of Web services required by the services described above. These include RDF parsing, storage and conversion of data to presentation formats. It is not necessary for the providers to host these services themselves but they must inform the consumer of where they can be found! Having completed these tasks, the providers may in fact have no 'running' services, but rather a series of files providing information both about the items in the collection and the services required to access them.

The consumer differs somewhat from the provider in that it is a pre-coded piece of software (known as an 'applet') that requires minimal configuration and need only be hosted at Web sites (typically those of collection providers). The chief advantage of this distribution process is that the actual user base (researchers, students, teachers, the general public etc.) can use the software without having to buy or download a 'shrinkwrapped' product – the software runs directly in the browser window. The consumer works primarily as the human interface, providing a visual control panel and display by means of which the user can search, navigate, compare items, create a personal collection, export to a presentation and, ultimately, resyndicate the data to make those collections browsable by other on-line parties.

Implementation

We made a number of strategic design decisions in implementing the code and tools discussed in the previous section. First, the project is entirely Open Source and is protected by the GNU Public License (GPL or Copyleft). GPL means that access to both compiled and source code as well as documentation is completely open to anyone (the code and documentation are stored with the SourceForge community and can be downloaded from http://vlma.sourceforge.net). The code may also be further developed by third parties, the only condition being that all derived software must also be distributed under GPL (http://www.gnu.org/copyleft/gpl.html). While the license does not officially ban the sale of such software, by ensuring its free availability it effectively prevents it. This is clearly of benefit to publicly funded institutions such as museums and archives for whom vendor lock-in can prove suddenly and catastrophically expensive. It also enables all interested parties to see how VLMA works, and this might encourage use and greater acceptance not only of VLMA but also of the concepts that support it.

Java (version 1.4.2), developed by Sun (http://java.sun.com/), has been used as the coding language for the consumer, as it is intended to be 'written once, run anywhere'. It has particularly strong features for creating programs which can be run from inside a Web browser ('applets'). In reality the use of Java is not as simple as it sounds, but a benefit to the VLMA consumer is that it can be run on all operating systems (patchy support for Java 1.4.2 on Mac OS X restricts the browsers it can use in that environment; it runs on Safari, which comes bundled with all Macintosh systems). The tools needed by the provider for RDF syndication are written in Perl, which also enable multi-platform implementation (http://www.perl.org/). Collections and services may be run on any platform and, indeed, may be distributed over several systems running different operating systems.

The object-oriented nature of Java also supports our third critical design decision, which was to make the code as modular as possible. The consumer element is built out of a number of modules that plug in to a central framework. This enables the applet to be as extensible as possible and encourages further development to provide alternative GUIs (Graphical User Interfaces) or consumer-run services rather than Web services in environments in which performance would be increased.

The modular components

Fig. 2: The modular components of the VLMA consumer

Summary

What then are the benefits of VLMA? One of the greatest advantages is that could reduce costs – it offers a completely free extension to any database, on any system, that makes a database instantly available to on-line users without risking the danger of being held hostage to proprietary software. In an age of limited public funds, this is no small consideration. Beyond having no installation costs, it adds value to the collection itself, providing services that enable users to explore not only items in the collection, but also the relationships those items have with others in collections the world over. As artifacts are rarely of great value in themselves but rather in what they can tell us of patterns in culture (material and otherwise), it is clear that the value of a multitude of collections is greater than the sum of its parts.

By increasing the ease with which users can access the data they seek, VLMA encourages them to make more use of it, both in terms of the quantity of items browsed and in the depth of information that they obtain about each artifact. This in turn will create greater motivation for collections and their funders to invest in documenting and publicising more of their collections.

Transparency is increased as all information relating to a specific object is linked to it directly and managed only by the collection owner. As a result, the danger of altered items masquerading as the original or metadata being arbitrarily withheld (other than by the provider) is lessened. This brings us to perhaps the most intellectually encouraging advantage of VLMA, its 'hands-off' approach to metadata. While we consider metadata models to be highly beneficial, it is increasingly difficult for providers to reach consensus on international standards for museum data and its presentation. VLMA works in parallel to such standards initiatives, providing a method by which collections implementing multiple standards can be browsed simultaneously, without exclusion.

Acknowledgements

VLMA is grateful for financial support from JISC in 2004-2005 and for institutional support from University of Reading (http://www.rdg.ac.uk/ure) and the Max Planck Institute for the History of Science, Berlin (http://www.mpiwg-berlin.mpg.de).

References

Berners-Lee, T., J. Hendler & O. Lassila (2001). The semantic Web. Scientific American. 284(5):34–43, May.

Crane, G., R. Chavez, A. Mahoney, T. Milbank, J. Rydberg-Cox , D. Smith, and C. Wulfman (2001). Drudgery and deep thought: Designing a digital library for the humanities. Communications of the Association for Computing Machinery 44(5), 35-40. Consulted 30 January 2005. http://www.perseus.tufts.edu/Articles/cacm2000.pdf

CIDOC CRM. (2005) CIDOC Conceptual Reference Model (CRM). A project of the Documentation Standards Working Group and CIDOC CRM SIG of the Documentation Committee (CIDOC) of the International Council of Museums (ICOM). http://cidoc.ics.forth.gr/index.html

Fensel, D., J. Hendler, H. Lieberman & W. Wahlster (Eds.) (2002). Weaving the Semantic Web. Cambridge: The MIT Press.

Hyvönen,E., M. Salminen, M. Junnila (2004). Annotation of Heterogeneous Database Content for the Semantic Web. Proceedings of the 4th International Workshop on Knowledge Markup and Semantic Annotation (SemAnnot), Hiroshima, Japan, Nov. 8. Consulted 30 January 2005. http://www.cs.helsinki.fi/u/eahyvone/publications/iswc2004/contentCreation.pdf

Lassila, O. and Swick, R.R. (Eds.) (1999). Resource Description Framework (RDF) Model and Syntax Specification. W3C Recommendation, February. Consulted 30 January 2005. http://www.w3.org/TR/1999/REC-rdf-syntax-19990222/

Priebe, T. and G. Pernul (2003), Towards Integrative Enterprise Knowledge Portals. In Proceedings of the Twelfth International Conference on Information and Knowledge Management (CIKM), New Orleans, LA, USA, November. Consulted 30 January 2005. http://www.priebe-it.de/en/../publications/PrPe03b.pdf

Smith, A.C. (2000). Disiecta Membra: Construction and Reconstruction in a Digital Catalog of Greek Sculpture. In J. Greenberg (Ed.) Metadata and Organizing Educational Resources on the Internet (Binghamton, NY: Haworth Press) 173-89.

Williams, D.J. (1983). Sophilos in the British Museum. J. Paul Getty Museum Occasional Papers 1, 34.

Cite as:

Fuchs, B., L. Isaksen, and A. Smith, The Virtual Lightbox for Museums and Archives: A Portlet Solution for Structured Data Reuse Across Distributed Visual Resources, in J. Trant and D. Bearman (eds.). Museums and the Web 2005: Proceedings, Toronto: Archives & Museum Informatics, published March 31, 2005 at http://www.archimuse.com/mw2005/papers/fuchs/fuchs.html