Demo
The following screenshots show how to use RaDON plug-in in NeOn Toolkit to repair and diagnose an ontology network:
Figure 1. Invoke the view of repair and diagnose ontology network. "Show View" dialog is invoked by choosing the menu of "Window" --> "Show View" --> "Other ...". From this dialog, one can find our view of "Repair and Diagnose Ontology Network" under the folder of "Radon".
Figure 2. Input ontologies and alignment (a). The items of "Ontology 1" and "Ontology 2" are corresponding to two source ontologies and the item of "Alignment" corresponds to the mappings between the two source ontologies. For our definition of inconsistency in the ontology network, we follow a global model semantics: An ontology network is inconsistent (resp. incoherent) if the ontology obtained by merging the ontologies and their mappings is inconsistent (resp. incoherent). The tab of "Ontology Network" shows the information about this merged ontology.
For the repair process we focus on repairing the
mappings, assuming the ontologies are already
individually coherent and consistent. Consequently only axioms in
the mappings could be removed to resolve incoherence or
inconsistency.
For each input, the details like how many axioms in an input ontology, how many unsatisfiable concepts in it, are shown in each tab of "Details of Input" separately. And the tab of "Namespaces" is used to indicate the relationships between the abbrivations and the corresponding namespaces.
Figure 3. Input ontologies and alignment (b). For the tab of "Alignment", we not only show the basic information mentioned in Figure 2, but also the mappings and their similarity values in the alignment.
If the alignment does not include mappings between the two source ontologies or it is in the format that we do not support, some messages will be displayed to users. Currently, we support the alignment format like INRIA format (example), Notation3 (example), Karlsruhe format (example) and Manheim format (example).
Figure 4. Debugging the ontology network. If the merged ontology is incoherent,
the system computes all the unsatisfiable concepts and displays them. When an unsatisfiable concept is selected, all the
minimal unsatisfiability-preserving subsets (MUPS or justifications) w.r.t. the concept are computed.
In a similar way, inconsistencies in the ontology network are debugged: For inconsistencies, all the minimal inconsistent subsets (MIS) are calculated and displayed in the section of "Handle inconsistency" in RaDON plug-in.
Figure 5. Repairing the ontology network manually. This step can be performed after we get the debugging results. Particularly, the minimal incoherence-preserving subsets
(MIPS) are obtained
according to the found MUPS if the ontology is incoherent. The found MIPS are presented to users for repair.
Each
MIPS is divided into two parts: The sets of axioms with the correspondences (which may be removed) and the sets of axioms that belong to the ontologies (which are assumed to be fixed). For the axioms in the removable part, their confidence
values are presented to the user to facilitate the selection of
axioms to remove. The unremovable part is also provided to make it
easier for the user to understand the contradictions.
If the ontology is inconsistent, all the minimal conflict subsets (MIS) with repect to this ontology will be computed and presented to users. One could repair the MIS one by one.
Figure 6. Repairing the ontology network automatically. The same information as that in the dialog to repair an ontology manually is shown. The difference is that users could not select the axioms they would like to remove. But we propose a set of axioms to be removed according to the confidence values of axioms in each removable subsets.
|
