The archaeological database—New relations?

Abstract

Over two decades have passed since the foundations of the relational data model were formalised (Codd 1970) and today a large number of Database Management Systems (DBMS) based on its principles are readily available. The better of these have attained a high degree of sophistication, running in a variety of environments — micros, workstations, minis and mainframes — and have achieved some standardisation through the adoption of Standard (or Structured) Query Language (SQL). As such, the user who invests much time in learning to use a DBMS and its development tools, for example INGRES, will have little problem when the present micro is dumped and a workstation appears on the desk. More importantly for archaeological information, the data, its structure, and application programs will also transfer with minimal upheaval. This is a salutary warning to those investing a great deal of resources in non-upwardly mobile micro-based DBMS and they are urged to consider employing either ORACLE or INGRES (the current flagships of the 4th generation language multi- environment relational DBMS) if they wish to ensure the longevity of their work. The reference to work rather than just to data is deliberate and the cornerstone of this paper, for information is not just data values; it is the context and meaning of those values that ultimately determine the usefulness of the data. Data structure, user interfaces, validation procedures, help systems and applications are inextricably linked with the raw data, giving it context and providing a crude but non-trivial 'knowledge base' without which data files may be useless, or even a negative resource, if misunderstood. Although high-quality relational DBMS did not come into general use as commercial products until the late 1980s, deficiencies in the relational model had already been noted in the previous decade. Important new products are likely to become generally available soon. Many of the major research areas of general DBMS have direct application in the management of archaeological data. The aim of this paper is to discuss some of the limitations and deficiencies of currently available relational DBMS, to review informally the most relevant areas of development (and one area which has yet to be developed), and to consider the implications for mainstream archaeology.