zhengj2007
commented
9 years ago From alanruttenberg@gmail.com on March 12, 2013 10:42:46
This sounds like a reasonable project. Note, however, that the reasoning performance will be impacted not by overall changes of an untemporalized version to an at-all-times version, but rather by the number of axioms we have added to generally firm up what the intended interpretations are, things like domain/range, local reflexivity, axioms connecting entities to regions, etc.
I'm not sure that we will be able to do this in a practical amount of time, given the other issues that need to be resolved, so we should discuss what kind of priority there is for this. My guess is that, unfortunately, the use it and report problems approach is the more practical one. In such cases I would look to see whether in the given case there are axioms that impact the performance but not the entailments. We could design subsets of BFO2 with weakened semantics for particular applications.
From cmung...@gmail.com on March 09, 2013 17:39:41
See Janna's email: https://groups.google.com/d/msg/bfo-devel/4iW_pVjfVIM/P7ZA_zBRXPcJ (perhaps Janna can provide original file)
If groups plan to adopt BFO2, they should be prepared rather, than trying to fix problems as they occur.
The BFO2 group should have some kind of sense of the reasoning performance issues we might face moving forward. For example, we should benchmark each ontology with and without temporalized relations. For the benchmarking a naive rel <=> rel-at-all-times mapping could be automatically performed on test versions of each ontology. This should be a responsibility of the BFO2 group, not individual ontology maintainers. We would then use the data generated to prepare recommendations for handling issues that arise.
Original issue: http://code.google.com/p/bfo/issues/detail?id=158