Tversky's Parameterized Similarity Ratio Model: A Basis for Semantic Relatedness

Numerous measures have been proposed to determine the semantic relatedness between words. Earlier approaches rely on the location of the words within the structure of a terminological ontology and are categorized into distance-based or information content models. More recently research has considered the overlapping of attributes or relationships, i.e., feature matching, between words or lexical concepts. This paper focuses on Tversky's prominent parameterized ratio model and its fundamental role in semantic relatedness measures. Existing semantic relatedness measures from the distance-based and information content categories are unified through this model. Through appropriate feature selection of properties of word objects in terminological ontologies, the model may also be used as the basis for proposing numerous other approaches to creating a variety of semantic relatedness measures