BACKGROUND:
‘Automatic webmapping’ in current systems means that the maps are generated from the spatio-temporal data by the system “working by itself with little or no direct human control”. This automation at present does not include the cartographic decisions as to what type of map to use for different data–types and data–instances. The link between data type and visualisation has to be made by a human, setting up the appropriate configuration parameters beforehand.

Fully automated mapping from data, with cartographic design decisions included, remains an interesting research challenge. It is clear we need to move away from the traditional way of working in web-mapping (using WMS) which leads to a sub-optimal combination of arbitrary map layers. Instead we need integrated mapping of data layers (see figure 3). There seem to be several approaches for this problem, and a promising one is the use of a formal map specification language. This would be a formalised specification (in the computer science sense) that defines the desired outcome using a declarative language. This language should have defined degrees of freedom (e.g. ranges of acceptable values) that allow for map creation in a controlled and consistent manner. It could be used by a service compiler that creates a service configuration file (e.g. an SLD) based on the formal map specification, plus (meta–)data and user input. The possible set-up is sketched in figure 2.

THE USE CASE AND THE RESEARCH:
The existing experimental 3rd edition of the National Atlas of the Netherlands serves as a use case.

The student needs to do a thorough literature study into formal specification languages. Research should be done into the strong and weak points of the specification languages and determine their usefulness for practical implementation in the use case: relatively simple thematic maps at the national (Dutch) scale. The best candidate identified should be used to do testing for the use case: How well  can we describe webmaps of the National Atlas using the formal language? Can a proof-of-concept implementation be made that automatically (re)creates the webmaps using the available datasources in combination with the formal map specification? The outcomes should be used to make recommendations for further development of the formal language.