Parsing SBVR-based controlled languages

Abstract

Conceptual schemas (CS) are core elements of information systems knowledge. A challenging issue in the management processes is to allow decision makers, such as business people, to directly define and refine their schemas using a pseudo-natural language. The recently published Semantics for Business Vocabulary and Rules (SBVR) is a good candidate for an intermediate layer: it offers an abstract syntax able to express a CS, as well as a concrete syntax based on structured English. In this article, we propose an original method for extracting a SBVR terminal model out of a controlled English text and then transform it into a UML class diagram. We describe a model-driven engineering approach in which constraint-programming based search is combined with model transformation. The use of an advanced resolution technique (configuration) as an operation on models allows for non-deterministic parsing and language flexibility. In addition to the theoretical results, preliminary experiments on a running example are provided. © 2009 Springer Berlin Heidelberg.