Candidate measures are identified with a clear link to each selected information need, for which the measurement information model of Annex A serves; new measures, possibly adaptations of existing ones, are defined in enough detail to allow a selection decision. Measures are selected from the candidates according to the priority of the needs and criteria such as those in Annex C, with the context information needed to interpret or normalise them, such as the programming language behind a lines-of-code count or the system attributes behind a requirements count. Each selected measure is fully specified: name, unit, a formal definition of how values are computed (with the input measures and constants of any derived measure), the collection, storage and analysis method, and the information need it serves; some definitions may already exist in the experience base.
The graph holds this control, the 0 it maps to, and the evidence behind each claim, over MCP and REST.