Wednesday, September 15, 2004

Something I forgot to mention earlier was consistency checking. The DL I've been translating describes quite a few rules for determining if a system is inconsistent. These are just as easy to translate into iTQL as any other rule, only I don't know what to do with the results.

In all other cases, the results become statements which get inserted into the datastore as inferred data. It is not appropriate (though entirely possible) to treat consistency checks in the same way. If I did then I'd need to define a new "inconsistent" class, which would be very messy, and could be easily missed.

I believe that inconsistency should be reported as some kind of exception, but I don't really know how and when to do it. It will probably depend on how we integrate the rules engine into Kowari. I'll have to ask AN's opinion in the morning.

No comments: