At the beginning o f the Nineties there was still a need to convince the archival community that in the centre of every archive's automation project must be a database and that the automation of main archival functions like description and retrieval must be based on a database management system12.
One of the main concerns for archivists in the use of databases in the early 1990s concerned the disadvantages linked to the shortness of text fields and the limited possibilities of operations allowed in memo fields. Many systems could work only in structured fields where free text was required or in a controlled language where natural language would have been preferable. For these reasons many archives chose information retrieval systems instead of databases. Then, with the growth of hardware standards and with a wider diffusion of Windows databases, these limits were overtaken. Further developments occurred, following the impetus to acquire images and sounds as data brought about better management of memory fields and led to more user friendly databases. However, many archivists found that most of the databases proposed by vendors were initially designed for libraries and for museums. This posed some major problems to archivists because libraries' databases and therefore their catalogues included units, all o f which were of the
same level and which had very little to do with the hierarchical structures of archival inventories. Museums databases were thought more appropriate for archives' needs since museums and archives have much more in common: origin and context are very important aspects, and the objects described are unique in both institutions.13
Databases have been considered for many years as the best way to incorporate pre-existent information into a structured system with clearly divided fields; either by means of insertion into the text of existing finding aids, of symbols or tags; or by encoding word-processed files; or by automated means such as OCR (Optical Character Recognition). Although structured databases encouraged exchanges with other organisations and the constitution o f collective databases, both on national and international level, they implied also high costs especially with pre existent descriptions which had to be converted and sometimes even manually entered.
Nowadays, database management systems correspond to the wider idea of archival description as a "dynamic"14 process and as an accumulation of information. This implies that information can be stored and amended or added to databases on a regular basis, constantly kept up to date and can accumulate with other data to allow more extensive and better directed searching on all informational elements about archives. Databases can also improve the ways in which information may be manipulated, merged and presented to users. For these reasons databases have also been considered as a single integrated finding aid15 or as the potential source of a number of different kinds of finding aid (which needs only to be printed
12 Green, A., Mise au point d'une strategic et de plans pour l'informatisation d'un service d'archives: Etude RAMP. Paris, UNESCO, 1991, 36
13 ibid, 37
14 International Council on Archives (ICA), ISADfGkGeneral International Standard Archival Description Adopted by the Ad Hoc Committee on Descriptive Standards. Stockholm, Sweden, 19-22 September 1999. Madrid 2000. Introduction, 11
out in whole or in part if there is a demand) all stemming from the same information input.
In recent years records management systems packages have been increasingly developed for archivists. Although it is no longer customary to regard archival description as a process wholly distinct from the control of information about the same archives for administrative and management purposes, some experts argue that "the coup de grace to traditional files based rather than system based records management has been delivered by the database management system which by definition manages information in more than one file"16. In fact, intellectual control involves a hierarchy of archival levels and it may be difficult or even impossible to adapt a computerised control system operating at a single level to the multiple levels required in archives work. Furthermore, records management packages are usually designed to control records at the item level, both documents and files, and may not be able to deal with records at higher levels of description and reproducing different systems of arrangement.
In the last decade, databases have definitely become a reality in the archives world and in the archivists' daily work. Archives are some of the organisations that have most benefited from developments in database technology and concepts especially in computer software with the capability to store data in an integrated, structured format, which enables users to retrieve, manipulate and manage data. Actually, many elements of archival finding aids require a presentation of their contents in a structured form. Such structured elements o f information can be readily divided and can therefore form fields in a database. Furthermore, the need
15 Kitching, C., The Impact o f Computerization on Archival Finding Aids: A Ramp Study. PG1- 91/WS/16, Paris, UNESCO, 1991, 7-8
16 Bearman, D., Electronic Evidence. Strategies for Managing Records in Contemporary Organizations. Pittsburgh, Archives and Museums Informatics,, 1994, 103
and motivation to be ISAD(G) compliant have led many archives to rethink their pre-existent databases or to merge many different supports into an "integrated system"17, i.e. a system which can handle information about all the holdings irrespectively of the storage medium and include management functions. Finally, database management systems have enhanced the ability of archivists to explain provenance and authority in relation to each fonds, concepts which although central for archival description are still little understood by end users.
Therefore, in order to better understand the main problems in the implementation of archival principles and practices in databases, with a major emphasis on archival description, it is necessary to give a brief overview and an historical excursus on the conceptual development of database design. There are three main periods or stages in the evolution of databases from the hierarchical databases in the 1960s through the relational databases of the 1980s to the object- oriented databases of the 1990s.
Hierarchical databases (1960-1980)
In a hierarchical scheme, all entities are connected through a parent-child relationship:
- each parent-record can have one or more child-records - each child-record can have only one parent-record.
Such a scheme can contain several levels, so that the final data model is a tree structure. The record type at the top of the tree is usually known as the root. A
17 Kitching C., The Impact o f Computerization on Archival Finding Aids: A Ramp Study. PGI- 91/WS/16, Paris, UNESCO, 1991,10
hierarchical database contains several types of records and links connecting occurrences of these records. It is fundamental to the hierarchical view of data that any given record-occurrence takes on its full significance only when seen in context (e.g. in relation to its superior). The main advantage of this kind of database is fast access time and the fact that hierarchies are the natural way to model truly hierarchical structures from the real world (e.g. departments and employees). The main disadvantage centres on retrieval, because even genuine hierarchical situations tend to develop into more complex many-to-many situations over time. The disadvantages o f this model are essentially the lack of theoretical base (no standard ways for definition and implementation of the database), the extreme complexity (one needs to be an expert to design these databases) and the lack of flexibility (insert, delete and update functions were very complex).
Relational databases (1980-...)
The specific structure of relational databases, embodied in tables and the relationship between tables, are aspects of the arrangement of electronic records, which reflect the use of data in an operating environment18. In relational databases the logical structure refers to the way the data within a record is organised, while the conceptual structure refers to the way the data is presented to users o f a record.
The original concepts of relational database systems were developed in the USA by Dr. E. F. Codd19 during the late Sixties and early Seventies and resulted in the first working implementation called “System R”, as a research project by IBM
1X
Thibodeau, K. "To Be Or Not To Be: Archive Methods for Electronic Records" Archival Management o f Electronic Records, Archives and Museums Informatics Technical Report, No.
13(1991), edited by David Bearman
during the seventies. It was based on the data Language SEQUEL, later renamed SQL. The Structured Query Language (SQL) became later on the standard for relational databases and it is recognised as such by the American National Standards Institute (ANSI) and by the International Standard Organisation (ISO).
The main concepts on which the relational model is based are these: All data are represented in tables.
The relational model is a way of looking at data, a prescription for how data can be represented and manipulated. This prescription is composed of three elements: a structural element, an integrity element and a manipulative element. The integrity element says that every relation (i.e. table) should have a unique primary key to identify entries or rows. The manipulative element of the relational model consists of the algebraic operators (select, project, joins, etc.) that transform relations into relations (and hence tables into tables).
The relational model is based on relational algebra.
A relational DBMS (Database Management System) uses the values in the data fields themselves to relate data items, rather than using physical pointers or indices. A relationship can be made dynamically between two data items. Using a traditional database data can only be retrieved using predefined access paths.
A relational database is set (multi-record) oriented, whereas a traditional DBMS deals with data one record at time.
The user never needs to be concerned about the physical location of data: there is an automatic navigation.
The user has only to express what information he wants and not how to find it.
The advantages of the relational database are the flexibility, ease of implementation and data independency.
Object-Oriented Database Systems (T990-..T
In a traditional relational DBMS, data is stored in tables that are accessible and joinable using primary and secondary keys. This means that the user or the program determines relations between data. As a consequence, the chance of having inconsistent relations is very big. In contrast, the Entity/Relationship-Model stores the relations in the data model and not in the programs.
Object-oriented database technology is based on the use of persistent object- combinations o f data and instructions that represent an action or an entity. In contrast, relational DBMS's data structure in rows and columns require relatively complex instructions to manipulate data. The terms “object base” and “object- oriented DBMS” are used to describe a class of programming systems with the capability of a DBMS, and with a combined Data Manipulation Language with the following features:
Complex Objects, i.e. the ability to define data types with a nested structure. Encapsulation, the ability to define procedures applying only to objects of a particular type and the ability to require that all access to those objects is via application o f one of those procedures.
Object identity, the ability of the system to distinguish between two objects that “look” the same.