The RGZM web databases

Java implementations on cultural heritage websites

All examples visitable at

Project Description
Technical Description
Tree Controls
Static Image Maps
Dynamic Image Maps
Other Java enabled Projects
Dating solutions
CAD driven

Project description

It all began in 1996. In order to create an image database on ancient shipping in Europe, eight European museums and institutes of ship archaeology decided to create a WWW based image database of more than 100 ancient ships known in Europe. NAVIS I was a project supported by the European Commission Directorate General X , Information, Communication, Culture, Audiovisual Culture and Audiovisual Policy.

The coordination of the different projects were done by the Römisch-Germanisches Zentralmuseum in Mainz (Tomba - Dr. Egg; Tangmausoleen - Dr. Greiff) and the Museum für Antike Schiffahrt in Mainz (NAVIS I-II - Dr. Pferdehirt) in Germany.

The central part of the first database NAVIS I was queryable in 8 European languages on its historical, technical and regional aspects. However, due to the fact that in 1996 the Unicode standard was not yet implemented in nearly browsers when the project was finished, local language coding settings and local fonts have been used for each language. In practice, this means switching ones browser settings for each language being used.

NAVIS I multiple language query mask


Due to the success of this shipwreck database, it was decided in 1999 to launch a second database - NAVIS II -, focusing at ship depictions on ancient monuments. The NAVIS I-partners were extended by participants from Poland, Tunesia and Israel. The language issue could be solved by the persistent usage of Unicode in JavaScript and Java modules, which started to be implemented in the major browsers around 1998/1999. Thus, by using only one client browser setting, all languages, including full Eastern Europe and bidirectional Hebrew, could be implemented. The implementation of a right to left explorer-like navigation tree was also successfull.

NAVIS II startpage
Bidirectional Hebrew database output

Technical description

Based on an MSAccess database and WindowsNT IIS, most standard query menus could be realised without huge problems and very low cost profiles. One of the mostly overlooked features of MSAccess is its ability to work with replicas, thus enabling multiple persons to enter data also offline all over Europe.

However, due to the complexity of the site, some Java controls proved to be most helpful:

Tree controls

The source code of such Java tree controls and compilers are available for free on the Internet. However, a growing tendency towards commercial packages like JBuilder or Visual Cafe enabling drag & drop creation can be observed, leaving the designer with no knowledge of the Java programming language alone when the insertion of a non-standard image or feature causes fatal errors, thus demonstating that doing serious work in Java requires access to and understanding of the source code.

For the purpose of the website NAVIS II, there was no tree control with multiple choice functionality available on the market. This was then developed in JavaScript. As with the Java controll based on class controls, JavaScript can also be generated dynamically out of any database underneath. The only disadvantage of JavaScript running in the runtime environment of the browser, appeared to be the changing or badly implemented DOM's by the different browsers available and thus makes it a vulnerable technology. The Class control based Java modules appeared to have the disadvantage, that major vendors like Microsoft since 2002 are not automatically supporting such virtual machines of non-Microsoft companies anymore.

Static Image Maps

Static Java based image maps have the huge advantage that all available information is loaded at one time, avoiding requerying the database for single question. However, no cross queries on combinations of features are realisable this way. Another disadvantage is that the map information is generally loaded from an external static file, not directly connected to the database, letting such projects end up in keeping not only the core database up to date, but also paying continuous time consuming attention to such a decentralised file organisation model. These static modules can also be used as kiosk presentations.

Static Image maps

Dynamically generated zoomable Distribution Maps

The dynamically generated web distribution maps in NAVIS I have the fascinating charm of highlightning only those features yielded by any dynamic database query, thus sending only the relevant feature information over the network. The map is directly connected to the database by ODBC or OLEDB, thus also allowing to make the highlighted findspots clickable as database drilldowns as well. The speed of this procedure highly depends on which web database querying technique is being used. Most professional programs such as InterDev or ColdFusion are allowing access to manipulate Recordcount values, whereas cheaper software such as Frontpage may bring you at first glance quick and nice looking results, but giving you no access to such essential SQL database querying features. All these very powerful tools, however, will not free you from the need to organize your data very well. It is rather pointless to buy expensive tools to increase database querying speed by 10 %, if the bad database design causes a 200 % loss in performance.

A further stage in the project NAVIS II was implementing dynamic map zooming functionality. This was also done by using and modifying freely available Java class files, resulting in a spectecular functionality, when combined with customizable plottable shapes. Each plotted shape can have its own popup lable, which is then clickable on its own.

Zoomable distribution map with popups.
Zoomed distribution map.

Other Java enabled projects

Further improvements, which have been implemented in other Mainz based projects, are Java based as well: Within the TOMBA project, a database on bronze age elite tombs with enhanced Java based querying techniques has been developed: Structurized keyword lists to query with, which can be dynamically generated out of the database, thus allowing better navigation and ease of maintenance.

Structured keyword list with time slided
Structured keyword list query output
Structured keyword list full image output

TOMBA dating solutions

- *working* (instead of theoretical) solutions for asynchronous archaeological time periods. This allows the user to compare finds from a cultural period familiar to him with other dating systems all over Europe
This can be done with any simple and cheap relational database like DBase, MSAccess or whatever:

- create a bunch of time phases which you think is appropriate for the material you are dealing with. In the worst case (Roman period), this might be phases of 10 years. Then make a list of culture periods and attribute them in a separate incidences table to each relevant phase (thus you can modify the time span of a cultural period when research makes this necessary). Then make a second incidences table where each object is attributed to the relevant phases (a macro will help you to speed up this process). Thats all: the database will now know that it has to return also the relevant phases from e.g. the middle European Bronze Age.

For practical purposes, also a Java slider with the absolute dates has been added for the broad puplic, since only a very smal part of the site visitors will have an idea what "Late Helladic III C" is about (see screenshots below).

Asynchronous dating technique
Slider with absolute datings

CAD frontend driven databases

The implementation of dxf based image maps, allowing hyperlinking and database drilldowns starting from single objects and pane/zooming as well is one of the major advantages of implementing Java techniques like the highly on GIS-focused website Tang Dynasty Tombs.

Java GIS frontend based on database output
Java GIS based on database output
Database output from Java GIS frontend

The advantages

The advantages are clear: Especially the module of the dynamically generated distribution maps proved to be more than just a trendy and fancy addition: it allows scientific research on distribution patterns through the web. The Java code is platform independent, thus enabling even Macintosh or Unix users to use these modules. Other programming techniques like DHTML and JavaScript can do similar jobs, but are currently too much browser-dependent, forcing the programmer to keep track of x different program versions for x different browser types available on the net. Not to mention keeping track of all future versions.

the Total Cost of Ownership is unbeatable.


The Disadvantages

The disadvantages of the general use of Java code on such database sites should be mentioned as well:

Putting this in balance, the advantages of having a mix of the easy handling and above all standardised Microsoft products combined with the platform independence of Sun's Java, enabled us to finish this project within the requested time, keeping all options open for future development.

Allard Mees