Resolving the "Weak Status" of Weak Entity Types in Entity-Relationships Schemas

Entity Relationship schemas include weak entity types, whose entities are identified by their inter-relationships to other entities. During the translation of the EER schema into a logical database schema, the weak entity types are either translated into logical units of the database (a relation or a class), or are embedded as attributes of other logical units. Formal studies of EER schemas either ignore the presence of weak entity types, or simplify their dependency structure. The presence of weak entity types in an EER schema may be problematic: A weak entity may not be easy to identify because it may be related to other weak entities in various ways, thus causing problems in schema comprehension, as well as in mapping it to a logical database schema. We claim that the presence of weak entity types in an EER schema belongs to an intermediate design stage, but the final EER schema must not include such entity types. We introduce an algorithm for resolving the status of weak entity types, following user directions. If the directions are legal, the algorithm yields an EER schema without weak entity types. The advantage of our approach is twofold: First, the translation of an EER schema into a logical database schema can be fully automated. This is essential for upgrading the EER model to support full database management. Second, it enables a fully formal study of the EER model.