This paper addresses the issue of documenting an existing legacy database by mining out its characteristics and derive the corresponding entity-relationship model. We developed algorithms to identify candidate keys of all relations in the relational schema, to locate the occurrence of a given candidate key as foreign key in any existing relation, and to decide on the appropriate links (relationships) between the given relations. Based on the mentioned analysis, we draw a graph that corresponds to the entity-relationship diagram, and predicts all possible relationships between relations in the existing relational schema. Finally, we derive the cardinality of each link in the graph.
展开▼
