Querying a Polynomial Object-Relational Constraint Database in Model-Based Diagnosis

Abstract

Many papers related to Constraint Databases (CDBs) theories exist, including proposals that present frameworks for the treatment of constraints as a new data type. Our proposal presents a new way of storing and manipulating constraints as a usual data, and of making queries about the constraint variables derived from an Object-Relational Constraint Database (ORCDB). In this work, the constraints stored in an ORCDB are only polynomial equality constraints. The proposal is based on Gr¨obner bases, constraint consistency and constraint optimisation techniques. Most works in CDB use spatial-temporal data as a case study, however this work presents an emergent engineering domain, that of fault diagnosis.