Problems and methods in database descriptions

One result from the current diversity of databases is that also the database descriptions are inconsistent. Because of the lack of commonly acknowledged guidelines for database descriptions, each producer of user documents (printed manuals, online help texts, software products etc.) presents the information in lapis own way. To aid the producing of better database descriptions, a research project for developing a general, systematical framework for database descriptions was starred at the Information Service of the Technical Research Centre of Finland (VTT). The project "Principles and methods for descriptions of database structures - DATABAS" was mainly funded by the Nordic Council for Scientific Information and Research Libraries (NORDINFO). An essential requirement for the database description is that searcher's needs have to be taken into consideration. The user documents should satisfy these information needs as well as possible. The situ a lions wile re database descriptions are needed can be divided into two basic categories, namely database selection and search profile formulation. When selecting an appropriate database, a searcher needs information about the contents of databases and how the material in a particular database differs from that of the other databases. Most problems are presently caused by poorly designed indexing languages. On the search stage, a searcher needs information about the features specific to a selected database in order to conduct the search as efficiently as possible. The most common problems ore that no data sleets exist, they are not comprehensive, or data is not presented in a uniform manner. The key issue is how to segment the information into appropriate pieces of data and represent them in salt a manner that any piece of information, such as language or document types, can be easily accessed. A list of such description segments was produced. Each description segment is an independent entity which can be accessed separately, but alley all together form a general "masters file, which represenls the characteristics of a database. According to this model, instruction sheers Ior describing a database directory entry and a dale sheet were produced. For testing purposes, seven different databases were selected and described according to lily suggested guidelines.