Introduction

The Documentation Project and the Museum Project are cooperative projects between the four universities in Norway. Since 1991, these projects have performed retro-conversion and digitization of analogue archives, books, images and other media types. During the building up of our diverse collections, the need for a model expressing relationships between place names has been strongly felt, both in the modelling and the presentation of each collection and in the present work towards deeper integration between the various collections. In our paper, we will present an object-oriented model for expressing relations between geographical objects applied on Norwegian geographical information taken from our digital collections, followed by a discussion about further development and implementation of the model.

Previous efforts

In connection with geographical reference systems for digital libraries we commonly find the use of UTM coordinates in the specification of geographical entities, as seen e.g. in the Alexandria project (Beard 1997). Whereas this method gives many possibilities both for use in single collections and in the interconnection of heterogeneous collections of information object, this method is not suitable for collections that contains a large number of references to geographical entities for which it is impossible or unfeasible to enter this type of coordinates. Whereas the use of UTM coordinates will be an important part of our system, we have to design a system which does not depend on this type of information. As the basic structure in the organization of the geographical information objects, we use the political geography of Norway. At any single point in history, this structure takes the form of a tree, as seen in fig. 1. However, this tree structured model is too restricted to cover collections spanning a time period. It have to be expanded, as we will see below. Vegard Elvestrand has published a standard for geographical classification (Elvestrand 1977) with municipalities as the smallest unit. The number of municipalities in Norway has changed a great deal through splits and unifications. His model is effective for the classification of literature; one of our collections is classified according to his standard (Eide, 1998 p. 287-288), but the underlying model is also too restricted for our use. His model contains classification symbols for every municipality that has existed, but it has no linking showing the actual changes. Norsk samfunnsvitenskapelig datatjeneste (Norwegian Social Science Data Services, NSD) has designed various databases based on the administrative structure of Norway, directed towards the needs of the social sciences to present data on present and historical statistics (NSD 1997, chapter 3). Their model is effective for historical models of municipalities, but it does not allow us to include the multiple systems of geographical information needed for cultural and natural historical information systems, as we will see below. The model described here is conform with the CIDOC object-oriented Conceptual Reference Model (CIDOC 1999), and we will follow the further development of the CIDOC model closely as we develop and implement our model of geographical information.

Places vs. place names

There are several other structures interconnected with the simple structure of fig. 1, e.g.:

• Other administrative regions, e.g. police districts, clerical districts
• Named geographical formations, e.g. lakes, mountains, rivers
• Property structures, e.g. farms, common land

In the geographical model in fig. 1, there is no distinction between a place and a place name. The difference between a geographical unity in the real world and the name we use to refer to this unity must be expressed in any geographical model. There are two main reasons for this:

• The same word often refer to several distinct places. One example of this is a municipality called Våler, which is found in the county of Hedmark as well as in the county of Østfold.
• Political structures may change. The area of the town of Hamar is larger today than it was ten years ago, whereas the municipality of Vang no longer exists (in a bureaucratic sense of the word exist), as it is included in the expanded town of Hamar.

In fig. 2, our basic geographic model is sketched, populated with several geographical objects. The idea behind the model is not to represent all potential information about geographical entities and their relations, but to include enough information to perform computer search, browse and interconnection between objects in an effective way. Each object in our model may have an attribute showing the time span that the geographical entity was known in that form. This does not mean that the object is placed in a time continuum, as it has a distinct, everlasting geographical location, which may be expressed in geographical coordinates. The time span is just an attribute showing which time period this specific geographical object was referred to. This is especially important for legal entities such as municipalities. Municipalities described as old or new in the figure have been split up or united. The direction of the graph is always from the bigger unit to the smaller unit: a unit pointed to is a part of the unit pointing from. To make this possible, we have split geographical entities covering parts of several other entities into parts shown on the figure as hp for hill parts and sp for sea parts. Each such part is part of both the sea as a whole and the municipality that this part is located in. The known place object contains a name, an optional time span, and references to other objects forming the graph. As some of our collections contain UTM data, and as we will be able to import such data for all our municipalities from e.g. NSD, the object also contains an optional UTM attribute.

Expansion into the unknown

We define each actual mentioning of names in texts and other media as an observation. In this context, observations are all references to geographical entities in information objects. Their relation to the model is seen in fig. 3. These observations are external to the model; whereas the model shows the objective reality (in the meaning agreed upon, checked), observations are subjective references made by a human being at a specific time, although this time may be uncertain or unknown. An observation will never refer to more than one place-name by definition. If there is more than one link from an observation to known facts, such a split expresses an insecure fact, in which case an observation may relate to one out of several possible known places, but we do not know which one. There is no limitations as to what kind of entities may be included in an implementation of this model, but we will have to decide on standards when we use the model in actual work. The development of such standards will be be discussed in our paper, together with search facilities and matters of implementation.