Relationships for dynamic data types in RSQL

Abstract

Currently, there is a mismatch between the conceptual model of an information system and its implementation in a database management system (DBMS). Most of the conceptual modeling languages relate their conceptual entities with relationships, but relational database management systems solely rely on the notion of relations to model both, entities and relationships. To make things worse, real world objects are not static as assumed in such modeling languages, but change over time. Thus, modeling languages were enriched to model those scenarios, as well. However, mapping these models onto relational databases requires the use of object-relational mapping engines, which in turn hide the semantics of the conceptual model from the DBMS. Consequently, traditional relational database systems cannot directly ensure specific consistency constraints and thus lose their meaning as single point of truth for highly distributed information systems. To overcome these issues we have proposed RSQL, a data model and query language introducing role-based data structures in DBMSs. Despite the fact that RSQL is able to handle complex objects, it does not support relationships between those objects. Therefore, this work adds relationships to RSQL by augmenting the data model and extending its query language. As a result, this extension allows for the direct representation of conceptual models with complex objects and relationships in the DBMS. Thus, relationships can be directly addressed in queries and the DBMS automatically ensures relationship consistency constraints as well as cardinality. In sum, a DBMS equipped with the extended RSQL is apt for storing and querying conceptual models and thus regains its rightful position as the single point of truth for highly distributed information systems.