Issue 639: Core and Application Models
A proposal by Martin Doerr to distinguish among 'core' and 'application' models, and define ther interface as well as their management.
Post by Martin Doerr (4 May 2023)
From the beginning of the CRM development, we followed a clear distinction between "core model" and "application model", as has been, in similar form, be proposed by other authors as well.
The basic idea is that the wealth of phenomena , even for the domain of factual information of the historical sciencesto cannot be be modeled proactively, and will, in general, not be consistent. However, high-level "superproperties" can effectively "streamline" specializations under a common core, that enables to a high degree recall, i.e. query containment for questions across multiple resources that can return reasonably small result sets of most relevant facts available for the respective research. The latter may then be processed locally for further refinement, deductions and evaluations.
The managerial idea was that CRMbase and extensions managed by the SIG form such a core, which is continuously verified by mapping of relevent applications for its genericity and efficiency for global querying.
Any specific application that goes beyond a global aggregation of data should develop local application models. Only when enough experience with these application models has been collected, it should feed back into the models managed by the SIG for modification or enhancement. This experience building aims at maintaining the long-term robustness and effectiveness of all constructs in the core. Nevertheless, the aim to keep all SIG-managed extensions consistent and harmonized, poses an ultimate limit to the size these constructs can develop into.
To my understanding, a lot of friction in the past years has been created by a lack of common understanding of the difference between these levels. My recommendations to develop certain constructs locally have sometimes been misinterpreted as rejections or lack of interest.
Therefore, I propose to
"define difference between core models and application models, clarify the way they interface technically and epistemologically, and define the managerial procedures associated with the relationship between them".
This should also be a didactic goal.