Fuzzy Object Comparison and Its Application to a Self-adaptable Query Mechanism

This paper describes a generic strategy for “intelligently” querying hierarchical semantic networks whose objects are characterized by fuzzy attributes. The strategy is self-adaptable in the sense that the network will keep structuring itself (as long as the user queries the system) by adding more and more detail to an underlying semi-lattice built on top of a (fuzzy) object comparison partial order. Technical bounds are imposed on the fuzziness of the system by formally specifying the network and reasoning about it. In the one direction, fuzziness is welcome insofar as it increases the flexibility and discriminating power of the comparison mechanism. In the opposite direction, arbitrary fuzziness may be counterproductive because it will end up destroying the formal properties desirable for such a mechanism. An experimental illustration of the proposed strategy is presented which re-interprets Prieto-Diáz faceted (fuzzy) scheme for software-component classification and reuse.