The main objective of the Action was to advance the understanding and use of relational structures in applicable object domains. This included the following sub-objectives: * To study the semantic and syntactic aspects of relational structures arising from real world situations. * To investigate automated inference for relational systems, and, where possible or feasible, develop deductive systems which can be implemented into industrial applications such as diagnostic systems. * To develop non-invasive scaling methods for the prediction of relational data, and to compare and possibly integrate a nominal scaling approach with numerical methods such as fuzzy relations, Bayesian networks etc. * To study and enhance current methods of relational qualitative reasoning about physical systems. Semantic and syntactical aspects of relational structures have been extensively studied in the Action with a considerable degree of cooperation among the members. The evidence seen of “real world situations” is successful considering the starting point and available time. For instance, the work on voting systems stands out as a successful collaboration between most of the participants. The work on relational preferences has acted as a focal point for work in other areas, bringing together groups who might otherwise not have collaborated with each other. The main and important outcome of the Action was to focus on really new combinations of diverse disciplines. There was a high degree of interaction between individuals from different member states resulting in an extensive list of publications. In the future, relational structures will be pervasive in many application areas. This Action established a new and wide cooperating network that in future can make significant progress in unifying diverse existing approaches.
