Do ask, do tell: a policy for successful semantic integration.
That is, ask and allow others to tell how they identify their subjects.
It does not mean, ask and then tell others a solution, approach, etc. to identify their subjects. (Including FOL.)
Users should be enabled to know when they are talking about the same thing. Using their own vocabularies.
Teach a user to integrate information and they have learned a new skill.
Teach a user to call an expert and they have gained a new bill.
Semantic experts have enough to do without making ordinary vocabularies require expert maintenance.
But isn’t the really hard part the actual mapping between one person/group’s semantic/ontologic world and another’s? Is there any real way to automate that?
And between cultures? E.g., ask folk from different cultures, “What color is that object?”
Comment by Kirk Lowery — September 27, 2010 @ 4:50 am
Yes, the hard part is mapping between different ways to identify the same subject. Which can be across languages/cultures but is just as hard within a single language. See the Blair article on why full text searching gets about 20% of the relevant documents.
It is possible to improve that by explicit identification of subjects, using specific identifiers or specifying the subjects that identify a particular subject and creating mappings between those identifications.
There are also data mining techniques that can assist in a human user in making the determination that two or more references are to the same subject.
My point is that for all of its foibles, a user’s own terminology is often sufficient for such mappings. It will certainly make a mapping easier for a user to use.
Comment by Patrick Durusau — September 27, 2010 @ 6:15 am