At least with regard to Core-AAM and HTML-AAM, it would be good for the mapping tables to be structured the same as much as possible, and for the specific mapping details to be presented in the same way.

While I might be biased, I prefer the HTML-AAM approach where the mapping table column headers note the name of the AAPI, e.g. "MSAA + IAccessible2", over the Core-AAM where the column headers are more specific, e.g. "MSAA + IAccessible2 Role + Other IAccessible2 Features. 

I prefer the HTML-AAM approach because the particular AAPI features and properties (roles, control types, localized control types, object attributes, states, events, interfaces, etc.) being defined are, for the most part (it's not perfect and there are still improvements to be made) expressly identified in a clear and consistent fashion. For example, see the mappings for the HTML caption element [1].

There is a degree of inconsistency in how the AAPI roles and features are presented for the ARIA roles in the Core-AAM. Granted, there's some inconsistency in the HTML-AAM as well, but I think the more explicit identification of the feature, i.e. role or control type or control pattern or state, etc. is an approach that results in a clearer specification.

While I have my preference as noted above, my greater concern is that whatever  approach is taken is used consistently across both the HTML-AAM and Core-AAM, given the especially close relationship between the two specs.


[1] http://rawgit.com/w3c/aria/master/html-aam/html-aam.html#el-caption