Task31From Kolflow ProjectRevision as of 07:20, 7 February 2011 by Desmontils-e (Talk | contribs)
Title : Revision and Merging of Ontologies. Code : 3.1 Start Date : 2011-03-01 End Date : 2013-02-28
Description : The goal of this subtask of the task is to design two algorithms: one for ontology revision and one for ontology merging. Revising a knowledge base KB1 by a knowledge base KB2 implies that KB1 may be changed, but not KB2. This makes sense e.g., when KB1 describes some inaccurate knowledge about a domain that are contradicted by actual facts described in KB2. However, in many situations, there is no reason to give a preference neither to KB1 nor to KB2, for instance when KB1 and KB2 are given independently by two (human or computer) agents. In such a situation, the problem is how to merge KB1 and KB2. From the state of the art about revision and merging, and from our own experience, it seems that an algorithm for the latter should be easily designed from on an algorithm of the former (though this remains to be proven in the framework of description logics). Revision and merging are both based on an epistemic entrenchment. An open question is how can information about epistemic entrenchment be obtained. A possible answer may be found in the use of ontology history (see task 4): if a and b are two pieces of knowledge such that a is acknowledged by more users than b is, it may be interesting to consider that a is more entrenched than b. Deliverables :
|
