Paranymy: Enriching ontological knowledge in WordNets

This paper studies and explicates the rich conceptual relations among sister terms within the WordNet framework. We follow Huang et al. [1] and define those sister terms by a lexical semantic relation, named paranymy. Using paranymy, instead of the original indirect approach of defining the sister terms as words that have the same hypernym, enables WordNets to represent and enrich ontological knowledge. The familiar ontological problem of 'ISA overload' [2] can be solved by identifying and classifying conceptually salient groups among those sister terms. A set of paranyms are terms which are grouped together by one conceptual principle, which are evidenced by their linguistic behaviors. We believe that treating paranymy as a bona fide lexical semantic relation allows us to explicitly represent such classificatory information and enriches the ontological layer of WordNets. All rights are reserved.