Davy’s aim is to try to reconstruct a domain model from source code. For this, he has manually (!) constructed a reference model from two existing Java systems: OpenPM and Endeavour. He built the model by inspecting the UI. For instance, in the UI below, there is an iteration button on project, so there should be a relation between Project and Iteration in the domain model.
Subsequently, he compares this model with the model as extracting from the code. He found that a signiﬁcant part of the reference model can be reconstructed from the code and that the systems do not contain much junk.
Therefore, Davy concludes that domain knowledge is recoverable from “modern legacy” code and therefore domain model recovery could be a valuable component of a domain re-engineering process.
Pre-print is available.