The hermes harvest component (hermes.harvest.mmf) is a means to "feed" model repositories easily. Hence, it is used on real world models in order to find out what parts of these models would be good to be stored in a model library. Therfore, it analyzes a given canvas with respect to known parts in an existing model library and with respect to tightly "coupled / connected" areas. Then adjustments can be made to the proposed slices which can be stored into the library eventually. Mind that known parts which are connected to newly stored parts are linked in the model library as well.
The requirements led to an infrastructure that should enable the following three contepts:
The prototype realizes the concept induced requirements and offers:
The architecture below shows that a framework is behind the solution. Hence, it can work with different model libraries as long as a marker and a corresponding saver is realized. Moreover, the framework can offer different splitter because for different model types different algorithm might work better than others.
This video shows the basic functionality how models can be stored into a model library. The process takes "known" parts into account and links them to the newly stored.
This video shows how the evolution approach assesses model quality and offers a staged approach that guides model quality.
This video shows how models can be "applied", i.e., reused. They are loaded from a model library and inserted into the open editor. Several libraries and several editors are supported.