CollectioOrg System
An Open Hyperdocument System for Museum Collections

Introduction

Over the last few years, museums find themselves increasingly under pressure:

• economic pressure for more efficient and rational ways of operation

• public demand for access to the common cultural heritage over the Internet

• call for scientific and cultural European collaboration

To these ends,Content Management Systems and Digital Asset Management Systems systems have been proposed. Whereas the former would concentrate on the creation and management of digital resources, the latter instead would concentrate on the aspects of publishing (or selling) the digital assets over various channels to the public, to institutions, commercial content providers etc. [DC-TI2].

Making all or selected parts of the collection accessible over the Internet, allowing collaborative efforts, rendering the management more effective and cost-efficient, opening new ways of marketing etcetera would bring in fact numerous advantages, as has been proved in other areas like news, legal, financial, medical.

But the following problem areas are still blocking the wide-spread use of such systems:

1. Museums are confined to one vendor's proprietary application or application technology means that the data's future is linked to that particular vendor's, with all problems regarding long-time storage and even possible data loss.

2. Costly knowledge domain islands are created, where information is at one's disposal inside the restricted information system, but without the possibility of collaboration with other departments or museums, thereby giving away possible synergism effects.

3. The specific information structure of a museum collection is manifold and complex, and traditional data formats are almost always lossy, i.e. the digitalized data – due to the constraints of the traditional table/row RDBMS – do not contain 100% of all previously available information.

4. Finally, contextual text documentation which is related but not regarded direct part of a specific object is mostly left out in the digitalizing process.

To overcome these fundamental shortcomings, the CollectioOrg system introduces the Extensible Markup Language (XML) [XML10], which has been deployed in other business areas with great success, for use in an open hyperdocument system.

An Open Hyperdocument system for the Interconnection of Knowledge Domains was first proposed by Douglas C. Engelbart in 1990 [Engelbart1990] to overcome the problems plaguing the airplane industry. At the time, because of incompatible document types and the lack of common protocol for interchange, important documents could not be exchanged neither horizontally: between departments of the same contractor, nor vertically: e.g., between departments and external suppliers.

To achieve interoperability, he argued for the introduction of an open document type and a transparent mechanism for linkage to other documents or media objects. Every document would be referenced by a unique catalog number and could be searched, retrieved, transformed and re-used by other knowledge domains. The main characteristics of his system were: mixed-object documents; explicitly structured documents; view control of objects form, sequence and content; library system allowing referencing by catalog number; personal signature encryption; access control; every object being addressable; finally inter-linkage between hyperdocuments and other system data.

The basic hypermedia document type of the Internet – the (X)HTML document type – and its related linking mechanisms adress these issues only in part [Connolly2002]: in fact, structuring, view control, adressability and library system have been developed only in part or are lacking alltogether.

Whereas at the time of Engelbart's writing, the isolated knowledge domains he refers to were the airplane industry, today it's the museums' whose infrastructure is characterized by complex, proprietary and mostly incompatible systems. It is not uncommon to find different incompatible databases even inside the same museum structure.

In a way very similar to Engelbart's proposal, the CollectioOrg system, to achieve the widest possible interoperability, specifies:

1. A XML hyperdocument scheme for museum records used for internal production and for web applications, which combines the advantages of a document structure with a database-enabled format,

2. A mechanism of hyper-linking using URNs, combined with a basic definition of a web service for inter connectivity with other departments, other museums and for external providers

The CollectioOrg system is therefore not a program, or an application, but the definition of a framework: an Open Framework, because all specifications are openly published, every interested museum is invited to participate, all specifications can be used without restrictions, and, finally, all technologies used are those specified by the W3C or OASIS and therefore non-proprietary.

XML for the Data

XML, the eXtensible Markup Language, is in many ways the perfect means of transport for museum records:

XML is a W3C specification like, for example, HTML. Therefore there is no risk of being trapped in a proprietary technology with all its negative consequences: not only the data created with this technology, but all the applications built upon it can be used – and shared, for example with other departments or smaller museums – without encountering licensing problems.

XML is application-independent. That is, you don't need the original application who wrote the data to read and interpret it correctly later on another system. XML documents instead can be created, read and modified on any OS with any editor, either using simple (schema-aware) text editor, or more complex (and mostly commercial) editors with customized graphical user interfaces.

The XML specifications are surrounded with a rich application environment: XSL-T [XSLT10] and XSL-FO [XSL10] for transformations into HTML, RTF, PDF, or Text; XPath [XPATH10 XPATH20] and XQuery [XQUERY10] for accessing and searching documents, XML Schema [XMLSCHEMA0 XMLSCHEMA1 XMLSCHEMA2] for controlling the document structure and values, etc. All these specifications also W3C specifications, too.

Since XML documents are basically text files marked up in XML, the format is particularly suitable for long-time storage, because it is, like a Rolodex card, human readable and usable even when the applications which created the data don't exist anymore. The data format will outlive the applications which created the data.

Since XML isn't binary (machine-) code but basically a text file with added mark-up, it is portable across all existing OS without the need for conversions. In short time XML has thus become an international standard for data exchange, especially between different DB (even RDBMS).

XML and all its surrounding technologies are fully Unicode enabled, with the standard encoding being UTF-8. This allows not only the use of foreign language scripts including, e.g., Old Italic (   ), but facilitates also the transcription of epigraphical texts using special characters (‖, ⋮, 〈, 《, ⊂, ⊃, ⌈o⌉, ˋoˊ) for interpunction and combining glyphs (ḁb̥c̥, āb̄c̄, a̅b̅c̅, áb́ć, âb̂ĉ, a̵b̵c̵, ȃb̑c̑, a̲b̲c̲, a̶b̶c̶) denoting missing or illegible characters etc. The Unicode specification even allows to define Roman numerals as such (Ⅰ, Ⅴ, Ⅹ, Ⅼ, Ⅽ, Ⅾ, Ⅿ, ↀ, ↁ, ↂ).

Unlike data stored in RDBMS, XML is a self-describing format: that is, element and attribute names in XML are mostly self-explaining. With the advent of XML 1.1 [XML11], even for element names the full Unicode range is available. This makes XML data understandable even in those cases, when all accompanying documentation has been lost and only the data stream is preserved.

The XML Schema specification, Part II, has made XML Data-Type aware [XMLSCHEMA2]. Not only are a wide range of data types available out-of-the-box for specifying element and attribute values (date, time, integer, etc.), but more types can be customized: e.g., a custom scheme for museum inventory code like sc-[0-9]{4}.

XML is equally used for the storage of structured text, for example large documents divided into various sections, chapters, and paragraphs, thereby preserving all the hierarchical structure. This allows one to re-deploy smaller fragments of the original text document, which can now be reassembled in new contexts: for example, search result pages, but also printed museum catalogs built upon chosen XML data. Data structures (measurements etc.) and document structures (textual descriptions of restorations etc.) can be mixed freely side-by-side, within the same XML document.

Information in XML is grouped not in rows and tables like in flat conventional databases (one field for every value) but in a hierarchical or tree-like structures, where the values of the single nodes have their meaning according to their position inside the tree i.e. its context. This lets you create very flexible patterns which reflect the way humans organize information.

View capability was one of the Engelbarts original specifications. It can be achieved using the mechanisms provided by CSS and XSLT so that the same data can be presented in different views, for example an Outline View, which would give the basic structure only, with foldable sections and paragraphs.

Finally, using XMLSignature [XMLSIGNATURE] for digital signatures and XMLEncryption [XMLENCRYPTION] for encryption of parts of documents or whole documents, there are instruments at hand to protect XML documents from eavesdropping and secure them from tampering. Ownership and authorship can be ascertained during exchange of documents, e.g. between museums, and certain informations can be hidden, e.g. insurance details, addresses of private collections etc.

RelaxNG for the Logic

XML and its related technologies allow the separation of logic and data. That is, the data and document parts which make up the XML files (called instances) are physically separate from an abstract definition of its correct syntax and content (called schema).

The schema defines firstly the sequence of elements and element groups; secondly, their possible attributes and children; thirdly, the possible data types (text, number, date, boolean, integer etc.). Therefore, schema file contains all the system logic: i.e., it prescribes exactly how information has to be stored in order to give a syntactically correct document. It offers word lists and controls the data-input by allowing only valid elements and values.

Like XML itself, the schema definitions are written in non-proprietary formats, which makes them usable without any licensing questions, and guarantees their long-term value. Mostly used is the XML Schema Definition – the format developed by the W3C – but RelaxNG [RELAXNG], an OASIS specification and at the same time an ISO Standard, developed by James Clark, has gained significant ground lately: it is widely credited with being easier to write and to maintain, especially for larger projects.

A physically separated schema enables data input outside traditional museum systems: data can now be written – with its logic controlled – all over the museum departments, without the need for being connected to a central server. Personnel can work on private lap-tops which can be taken home at night – since there is no proprietary software, there is no risk of theft or breach of licensing. Just having the schema file/s on a floppy disk is enough for controlling all the data input, word lists included. This is in stark contrast to the data-input interfaces of traditional systems which have all the data input logic "hard-wired", i.e. the built-in definitions are proprietary code which cannot be separated from the system.

Identification and Reference of Resources

Modelling Relationships between Documents

Graph-based data models for modelling relationships between hyperdocuments have been analyzed by Eugene Eric Kim and Ken Holman in 2002 [KIM-HOLMAN2002]. They are used in two ways: the CollectioML Infoset provides a graph model for inter-relationships between documents, and an external graph modelling language (Topic Maps) provides for relationships between different classification systems.

Reference Platform

A Reference Platform has been built which is used as showcase and for experimental features alike. Nearly all components are built using Open Source software -- the only exception being Java which is used to achieve cross-platform capability.

The system is developed in the classical 3-tier architecture.

The persistence layer is a native XML database. A native XML database used as storage container (a so-called XML Repository) is the only efficient way for storing and retrieving XML documents, especially structured ones, because it uses the same range of W3C technologies as the rest of the system and is therefore easy to integrate.

XQuery, another non-proprietary W3C technology, is the tool for interrogating large collections of XML documents. XQuery not only allows for search and retrieval of XML nodes – including full-text search – but can be used to build new XML documents out of the retrieved XML nodes. This facilitates, for example, its use in search interfaces, where lists of search hits can easily be presented.

XUpdate is used for administration work across whole groups of documents, like changing values of elements or attributes or updating the XML document structure.

The logic layer of the system catches the incoming GET requests and interprets them according to his in-built rules: sending an XQuery to the database, de-referencing the URN, eventually connecting to outside resources, executing a series of transformations and finally representing the composite result. By hiding the inner operations we are able to have a consistent interface, regardless of what resources are tinged or how the results are composed.

The presentation layer has two basic modes implemented: List View and Document View. Like any search portal, a request first returns a List View – document records, image records, related documents, inventories – and then the chosen document.

With regard to visualization and export, the W3C technologies XSL-T and XSL-FO are perfect solutions. They have become widely adopted because of their power, transparency and ease of customization. In fact, a lot of templates – e.g. for DocBook – are already available. XSL-T is used for easy export to XHTML [XHTML11], also for export to plain text, such as the report format requested by the ICCD. Instead XSL-FO is the first step for transformations to RTF and PDF.

The interconnectivity between domains is offered along the principles of a REST architecture [Fielding2000]: a state-less transfer of representations of resources where each server is connected via HTTP and where the data stream is pure XML (XML-over-HTTP). This allows interconnection not only inside a museum domain, but also between different Internet servers.

Summary

Implementing the CollectioOrg system as a service that can be reached via LAN/WAN or between selected museums or even as public Web Service that can be accessed over normal Internet protocol has far-reaching conclusions:

1. The resources of formerly isolated knowledge domains (museums' information systems) are accessible and interchangeable:

   This allows not only a more efficient collaboration between different departments and museums but also between museums and universities or institutes, facilitating scientific research and allowing pan-European research projects. At the same time, the resources can be published to the general public, allowing wider participation and raising the public awareness of the museum.

2. The resources are able to create larger entities by transclusion:

   By aggregation of information resources from different knowledge domains it is possible to have a common generalist view over all resources. Here we are using the transclusion principle, i.e. not copying but referencing of external resources [Nelson1993]. One possible view collects all the pieces of a collection even when the collection itself is dispersed over various museums and countries. Other views comprehend collocations (e.g. all statues or objects inside a hall), or periods (e.g., all Roman busts from the Antonine period), or classifications (e.g., all Doric columns), or historic settings. The aggregation of resources is bigger than their sum.

3. The resources are uniquely addressable and have all the necessary characteristics of hyperdocuments:

   The use of URNs (instead of URLs) gives the museums' objects the characteristics of real objects which exist per se, and not only inside HTML pages describing them. Furthermore, this allows incoming and outgoing references not only to other documents but to any other kind of data – even data in foreign formats and applications. This is a precondition for building applications on top of the CollectioOrg system, e.g. Content Managing Systems for managing the objects, or Digital Asset Management Systems for publishing them over various channels.

4. The resources are in structured, open and internationally recognized formats, and therefore they are reusable for other applications:

   Standard data formats and standard access interfaces are the preconditions for building third-party applications, which can access data over the Internet as web services or via an internal API. For example, the development of Content Managing Systems is immensely facilitated by standard formats, and the same is true for Digital Asset Management Systems and the necessary implementation of a Digital Rights Management for granting differentiated access levels for partner institutions, commercial partners etc. (e.g. access to high resolution images).

Usage Cases

Uses of the CollectioOrg System go beyond a mere Museum Information System. The following examples – only few of the many possible – comprise external organizations and service providers, the scientific community and the general public:

1. Web Portals:

   The most common of all usages is the opening of a web portal for showcasing the museum inventory to a broader public.

2. Mobile Electronic Museum Guides:

   A mobile museum guide systems can be realized in the same manner, with the only difference being that the output is customized for the use on small hand-helds and only part of the available information is displayed.

3. Virtual Exhibitions:

   The combined net of federated servers allows exhibitions of material stored across different museums and different countries to be displayed in virtual showrooms. This can be used to gather art objects which are otherwise inaccessible, to reunite collections which are now dispersed, etc.

4. Catalog Publishing:

   The database can also be used to produce printed catalogues: either the collection's information is assembled as a list, to be handed over to the typesetter in native XML format (modern typesetting software can handle XML) or the database output is rendered directly in PDF or Postscript.

5. Museum Information Interchange:

   Museums can access another museum's collections on-line, and check their own items against theirs. This is especially useful for items which come from common historic collections and are dispersed across multiple museums.

6. Virtual Reality Projects:

   Using the HTTP connectivity, the museum material can be made available to Virtual Reality projects, even remote ones, to provide background information about the objects displayed.

7. External providers:

   External museum service providers can build their own applications on top of the museum's web service for the whole range of services they offer. This can be used, for example, by restoration and storage companies, or companies providing image scanning services.

8. Museum exhibition organizers:

   Museum exhibition organizers can keep track of the latest collocation of the items, and use their physical descriptions to organize transport and display more efficiently.

9. Providers of Digital Asset Management Systems:

   External DAMS providers can integrate their system into the museum's web service, eliminating the costs for a second, parallel web presence.

10. Insurance Companies and Police Departments:

    Insurance companies can be given special access control to the museum repository (with secure authorization) to check the insurance coverage and status of the museum objects. Police departments can use the system to look for items recovered or reported stolen.

11. Scientific Research:

    Queries – full-text queries or queries by category – across multiple servers are essential to research projects. Universities and research institutions can even include the URNs of the museum's objects inside their texts, allowing the reader to automatically look them up on the museum's website.

12. Cultural Heritage Management:

    The CollectioOrg system allows a common view over nationally and internationally dispersed museum collections and provides thus the necessary tools for the evaluation and management of the common cultural heritage.