Enhanced verbalization of ORM models

Abstract

Fact-oriented modeling approaches such as Object-Role Modeling (ORM) validate their models with domain experts by verbalizing the models in natural language, and by populating the relevant fact types with concrete examples. This paper extends previous work on verbalization of ORM models in a number of ways. Firstly, it considers some ways to better ensure that generated verbalizations are unambiguous, including occasional use of lengthier verbalizations that are tied more closely to the underlying logical form. Secondly, it provides improved verbalization patterns for common types of ORM constraints, such as uniqueness and mandatory role constraints. Thirdly, it provides an algorithm for verbalizing external uniqueness and frequency constraints over roles projected from join paths of arbitrary complexity. The paper also includes some discussion of how such verbalization enhancements were recently implemented in the Natural ORM Architect (NORMA) tool. © 2012 Springer-Verlag.