Author :
Kaufmann, Matt ; Portmann, E. ; Fathi, Madjid
Author_Institution :
Dept. of Inf., Univ. of Fribourg, Fribourg, Switzerland
Abstract :
Traditionally, ontologies describe knowledge representation in a denotational, formalized, and deductive way. In addition, in this paper, we propose a semiotic, inductive, and approximate approach to ontology creation. We define a conceptual framework, a semantics extraction algorithm, and a first proof of concept applying the algorithm to a small set of Wikipedia documents. Intended as an extension to the prevailing top-down ontologies, we introduce an inductive fuzzy grassroots ontology, which organizes itself organically from existing natural language Web content. Using inductive and approximate reasoning to reflect the natural way in which knowledge is processed, the ontology´s bottom-up build process creates emergent semantics learned from the Web. By this means, the ontology acts as a hub for computing with words described in natural language. For Web users, the structural semantics are visualized as inductive fuzzy cognitive maps, allowing an initial form of intelligence amplification. Eventually, we present an implementation of our inductive fuzzy grassroots ontology. Thus, this paper contributes an algorithm for the extraction of fuzzy grassroots ontologies from Web data by inductive fuzzy classification.
Keywords :
Internet; fuzzy reasoning; information retrieval; knowledge representation; natural language processing; ontologies (artificial intelligence); pattern classification; Web data; approximate reasoning; deductive reasoning; inductive fuzzy classification; inductive fuzzy cognitive maps; inductive fuzzy grassroot ontology; inductive reasoning; intelligence amplification; knowledge processing; knowledge representation; natural language Web content; semantic extraction algorithm; structural semantic visualization; Cognition; Data mining; Ontologies; Semantic Web; Semantics;