Formalism for Detecting Version Differences in Data Models

In the architecture/engineering/construction (AEC) industry, a large number of data models (e.g., data exchange standards and task-specific data models) have been created and utilized to represent and exchange data in software packages. To meet the ever-expanding requirements for modeling real world information, the data models need to be updated frequently. Accordingly, those who need to implement these data models in their AEC-related software which often requires that they possess civil engineering domain knowledge, have to change their existing implementations for compliance with these models to account for the latest update. Before adopting changes of such data models, those developers working at AEC-related software companies must precisely identify which parts of the data models have been modified in a new release. Given the growing scale and complexity of today’s data models involved in the AEC domain, identification of differences in two versions of a data model is a time-consuming and error-prone process, when performed manually. A semiautomated approach that identifies the differences in two versions of a data model could enable a rapid update of existing implementations of the model in AEC-related software. Due to the likelihood of having some commonality between the two versions of a model, it is possible to automatically identify version differences accurately. In this paper, we present an approach for detecting the differences between two releases of the same data model accurately and efficiently. This approach incorporates taxonomy for describing possible differences between two versions of a data model and provides a way to classify these differences. A prototype is implemented and used to validate the approach with the recent releases of some real world data models. The approach developed in this paper can help AEC-related software developers adopt and implement data models in their software systems.