As best I can tell, XBRL allows SEC filers to use the same concept for multiple facts so long as the context is different. I'm having a difficult time understanding how to include/exclude a fact for a given roleURI (a statement, for instance). I believe this ability is somehow related to context, but there doesn't seem to be an obvious link between the concepts asked for in the presentation document and the appropriate concept in the instance. To ask this a different way:
1) The company has several roleURIs (networks), perhaps one of which is "http://www.bigcompany.com/role/StatementOfIncome"
2) In the *_pre.xml document section related to this network, the company has asked that a "Revenues" concept be displayed.
3) The instance document has multiple "revenue" items, each with different contexts, and some with segments related to the company's sub-entities.
How do I determine that a revenue item with a certain context belongs on the StatementOfIncome roleURI, and another one should be excluded?
Thanks for any hints or resources...
As you note, the presentation linkbase is context-agnostic. From the exhibits themselves, there is no explicit way to know which numbers were actually printed on the Income Statement and which facts using the same element/concept are from other schedules.
The SEC uses programmatic means to determine what will be rendered where in its Pre-Viewer and Viewer; they explain some of that in their FAQs and Interpretations (see, for example, question B.3 at http://www.sec.gov/spotlight/xbrl/staff-interps.shtml). They look at key words and phrases.
The SEC makes the source code to their rendering engine available at http://www.sec.gov/spotlight/xbrl/viewers.shtml - you may be able to leverage it or find other hints of how they determine what to print and when.
XBRL was not designed to recreate the original presentation; later developments including Inline XBRL and the new Table Linkbase do more to preserve presentation placement.