Position dependent information services require effective location referencing

Claus Dorenbeck
TeleAtlas B.V.

Claus.dorenbeck@teleatlas.com


Introduction
Position relevant information services consist of two parts. The first part refers to the position, the second to the information. These parts are independent of each other. A position is usually defined as a set of co-ordinates. In this paper emphasise is given to the description of geographic phenomena. A geographic phenomenon has a location, which is a position as well as information about its type, and possibly names. A location can be a road, a bridge, a river. These locations may all have the same position in terms of co-ordinates. The information refers to something related to the geographic phenomenon. This may be a hotel with available rooms, a traffic jam, and the venue of an event.


Perception of space
Scientifically a geographic phenomenon can be described unambiguous and in great detail. "Downing Street 10" is a geographic phenomenon, most people understand as the location where the British prime minister lives. Does anyone know in which part of London or at what major crossing this is? Even if the full address would be given, most people will only understand, that "Downing Street 10" is somewhere in London. However, the majority of people living in London know very well where "Downing Street 10" is and even more important, they are able to relate this to their own location.

Apparently human beings have a different concept of space and the containing geographic phenomena than pure mathematics. From a mathematical point of view, the approximate co-ordinates of the house "Downing Street 10" are completely sufficient to know where this is and how this relates to anything else in space. 
The further our place of living is away from a described geographic phenomenon, the more difficulties we encounter in putting this into a sensible context of our movement. Everyone knows that the Eiffel Tower is in Paris. Does anyone know, how the location of the Eiffel Tower relates to the Airport, the city hall, any given place within Paris? Those of Paris know, those visiting Paris often know, those who see Paris only infrequently might have an "idea", but those who go there the first time will certainly need a city map.

In order to relate geographic phenomena to our current position, maps have been invented. A map is the agglomeration of geographic phenomena describing the space in which we live. A map allows us to find the geographic phenomenon we are at and understand relations to other geographic phenomena. 


Digital maps
In the early days, maps have been put on computer systems by simple scanning them. It became obvious, that this method allows only very limited computations upon the content of maps. These raster maps were just pictures, revealing nothing about their content to a computer system. For that reason, the content of maps has been structured into an appropriate data model. Each feature of a map is then digitised and assigned to the appropriate class of the data model. Additional information on the digitised geographic phenomenon is captured in a variety of attributes. Digital maps contain usually much more information than a paper map. A digital map does contain the complete legend, various drawing instructions, indices, alternative names and even functional classifications. The content is classified, much more information about each feature of a map is capture, e.g. a line represents a road with a name, a number, a functional road class, a form of way, and possible turn restrictions.

Currently digital maps suitable for traffic and travel information systems are standardised within ISO TC 204, WG 3. The name of this standardised data model for exchange purposes is Geographic Data Files (GDF). GDF describes an almost complete model of the world, emphasising traffic and travel needs. Several map vendors capture data according to this standard. However, some freedom of implementation of the GDF model is given to the data vendor. For that reason, a GDF map of vendor A may look slightly different then a map covering the same area from vendor B.


Position dependent information systems
One of the very first position dependant information systems is the Traffic Message Channel of the FM Radio Data System. The intention of this system is to convey traffic information via a digital FM sub-channel to motorists. At the time of its original development only very few navigation systems existed. For that reason, a decision was made to allow simple text based receivers to decode a given traffic message without utilisation of a digital map. Each location, i.e. all relevant street elements well known by motorists, received an authoritative number. This number referred to a key in a table. Within this table, one was able to find the name/number of the road, the general direction and finally the exit numbers/names in between something has happened. This approach allowed the coding of each "freeway/major highway" road element, i.e. part of road between any two exits/entries. With the given information, any motorist, regardless of his origin, was able to determine whether a given traffic message would have an effect on his journey. This approach works well for a structured highway network. It fails as soon as city roads, that are usually only known to locals, are being referenced, regardless how well the content of the table is prepared. Another draw back of the method has been encountered by map vendors who have to manually assign the above mentioned numbers to the appropriate road element within their digital maps. In order to overcame this obstacle, European map vendors started investigations on methods of map generic location referencing.


Map generic location referencing
In the late nineties, a group of ERTICO members started to develop a map generic location referencing method. The intention of the group was to find a method in describing the road network solely by using the content of digital maps, i.e. GDF maps. Since an alternative to the above mentioned TMC service was sought, the road network was of major concern. It was quickly understood, that co-ordinates alone will not serve the purpose. Each map vendor used different source material and had different techniques for digitisation. This results in slightly different co-ordinates. Since the road network is very dense in highly populated areas, a limitation to the relevant co-ordinates was not an option. In order to describe the road network, the decision was taken to use intersections. A pair of x and y co-ordinates as well as the names of the joining roads can describe each intersection. Some manual trails were successfully made and eventually an European project called EVIDENCE was launched in 1998. Within this project, the method was revisited, some rules on the selection of the co-ordinates and names were defined and coded in the various systems of the map vendors in order to perform practical tests. These tests revealed that the method works in general, however, in some 20% of all cases, it was not possible to say whether a intersection was interpreted properly or not.

Further research on the issue lead to a slightly different approach. Instead of focusing on the road network, a more intensive look on the data was taken. Even though each vendor had slightly different implementations of the GDF standard, a general similarity remains. The new approach considers an unambiguous classification of a digital map content according to the functionality of a given geographic phenomenon. This can usually be done in an automated way, since the functionality of a geographic phenomenon can be derived from the feature it represents or various values that are captured in an attribute. It has to be understood, that for this method, a class catalogue of geographic phenomena is needed.

The class catalogue of geographic phenomena will contain information on which co-ordinate of the according geographic phenomenon is taken as well as the rules on how to derive the relationship between a certain class and the given datum of a digital map. When a geographic phenomenon needs to be described, a digital map is taken, the according element, representing this geographic phenomenon, is selected and a code, consisting of a co-ordinate, a class and possibly some sort of abbreviated name/number is created. In the receiving system, the co-ordinate is used to determine the area interest, the class to determine the relevant elements and the abbreviated name to make the final selection. The code is no longer matched, but interpreted. This allows for different maps on the sending and receiving side.


The mobile environment
Within the anticipated mobile devices it is very unlikely to find digital maps. It might be possible, to find a navigation system within a vehicle that interacts with the mobile device. In this environment the above mentioned method will operate very well. However, a mobile user may want to relate a given position dependant information with his current location. In this scenario the position of the user needs to be determined and correlated to the location of the given information. A variety of issues may result in the fact, that only a specific service operator may be able to determine the users precise position. A user may have a preferred choice of a digital map for the presentation of any given information. It must be assumed that within the mobile environment a user may want to relate position dependant information to a web site containing the map of his choice. For this reason it must be possible to exchange location references between different web sites.  Assuming that an information for a certain position contains the appropriate location reference, any (web) site having access to a digital map can present this location in relation to the users current location. Since digital maps are structured, a wide variety of different details and generalisations can be presented. This allows the user to determine the level of presented detail, either according to his familiarity with the area, his distance to the location or whatever else is preferred. 


Conclusion
The perception of space or geographic phenomena is highly dependent on the distance of a geographic phenomenon to an individuals familiarity with the geographic area of daily activities. Resolving the issue of perception requires maps with various levels of detail. Digital maps offer the ability to present various levels of detail and content. A variety of digital maps of different vendors exist. An element of one digital map cannot be exactly matched with an element of another digital map, even if the same geographic phenomenon is described. Co-ordinates differ due to used source material, attributes of a geographic phenomenon may be differently modelled.  However, given a proper description of a digitised geographic phenomenon within one map allows another digital map to correctly interpret the geographic phenomenon. Only maps allow to understand the location of a position dependent information service and present it to the user in the desired context. Describing a geographic phenomenon in a digital map and interpreting it in another digital map is the key to effective location referencing.