Companies: A Combined Approach to
Information Retrieval from Legacy Databases
Martin Atkinson*,Ole Martin Winnem**
* Joint Research Center of the European Commission, Ispra, Italy.
** Sintef Telecom and Informatics, Trondhiem, Norway
Corporate experience of large industrial companies exists in many forms, most commonly, it is stored in databases that are located at different sites. These databases are often designed and constructed for specific industrial activities or domains, such as: design, engineering, manufacturing, marketing, operations and maintenance. There is often a strong correlation between the information stored in these databases, especially with respect to particular activities or product lines. However collaboration between different domain experts is necessary in order to extract pertinent information concerning a particular product or activity.
The objective of the Esprit Project Noemie was to automate as much as possible the access of information across industrial legacy databases in order to aid corporate decision making processes. This implied capturing the knowledge of domain experts as well as the domain specific constraints of users, whilst at the same time combining available software technologies for distributed data access, data correlation and knowledge retrieval.
The results from this project provide a methodology for domain modelling and problem solicitation, a software architecture supporting different data search, inference and correlation modules and a user interface supporting the methodology and tool configuration. These have been proved through configuration into two user applications.
The developed methodology was based on mixture of Common KADS ( Wielinga et al 1994) knowledge engineering Components of Expertise ( Steels,