Actions

Ontolog Forum

Remarks Extending the discussion in the Joint Communique to the UpperOntologySummit

Notes relating to numbered points in the UpperOntologySummit/ExtendedRemarks/JointCommuniqueDraft

  • Point (1)
    • Existing relational database systems and the associated SQL query language already support some elements of reasoning. Formal ontologies extend the reasoning capabilities of such systems, and axiomatized formal ontologies provide even more expressiveness and flexibility in the declarative representation of knowledge. Other forms of reasoning beyond those directly supported by current axiomatized ontologies may also be required for certain knowledge-based applications. The current initiative is primarily concerned with increasing the utility of the ontology component of knowledge-based reasoning systems.
    • The distinctive characteristic of "knowledge-based reasoning systems" is the generation of inferences on declaratively expressed knowledge, and the use of inferencing rules that are themselves declaratively expressed as part of the knowledge base. "Declaratively expressed" rules are those using the standard logical operators (or equivalents) with a set of "if-then" conditions, which are not embedded in procedural program code.
    • One description of automated reasoning can be found at:

http://plato.stanford.edu/entries/reasoning-automated/

  • Point (2)
    • A "Common Upper Ontology" is an upper ontology used in common by some community to achieve semantic interoperability. At any one time, different communities may use different Common Upper Ontologies. The panelists do not expect any agreement in the near future on one fully axiomatized upper ontology that will satisfy all potential users. Mechanisms to accommodate alternative theories, different viewpoints, possible worlds, and multiple contexts will be required in any Common Upper Ontology.
    • If agreement on use of a single Common Upper Ontology within a community is not feasible, a small number of upper ontologies that have detailed interrelations may also provide some practical level of interoperability.
  • Point (3)
    • Each of the existing upper ontologies represents a somewhat different perspective on the way knowledge can be organized, resulting in different structures at the top levels. The continued exploration of such alternative paradigms of knowledge representation can be expected to provide information on the relative benefits of each paradigm for different applications.
    • (Wide agreement - SRR)
  • Point (4)
    • The benefits of having some agreed set of commonly used concepts with representations that can be imported into any of the existing axiomatized upper ontologies will be to permit the largest community of developers of domain ontologies to use the common subset as a reference set of elements that can be used to specify the meanings of the elements in their domain ontologies. This will allow domain ontology builders to recognize when the intended meanings of the concept representations they use are identical to or different from those in other domain ontologies. This will reduce or eliminate the ambiguity in the meanings attached to ontology terms and permit different reasoning systems to arrive at the same inferences from the same data. The communities using any one existing upper ontology as ther Common Upper Ontology will be smaller, and have fewer of the networking benefits achieved by the larger user base of all of the ontologies together, but will be able to use all of the more detailed ontology elements and utility programs of that upper ontology, which may be required for certain applications.
  • Point (5)
    • A Common Subset Ontology would provide some of the functions of a Common Upper Ontology, with less detailed axiomatization and fewer of the more abstract elements. A goal will be to create a simplified ontology whose elements can be accurately imported or translated into elements compatible with all of the existing upper ontologies. This will permit prospective users of the upper ontologies to experiment with less complex representations that will be reusable in all of the upper ontologies when the capabilities of those systems are required. Depending on the size of the common subset that can be agreed on, it may also serve to enable semantic interoperability for reasoning tasks less complex than those supported by one of the existing Upper Ontologies.
    • (Weaker support - SRR)