Proximity measures for terms based on fuzzy neighborhoods in document sets Original Research Article

Two types of proximity measures between two terms are defined using a fuzzy neighborhood in a sequence of terms of a document set, where the concept of neighborhood is borrowed from generalized rough sets. These proximity measures are a refinement of similarity measures based on the term-document cooccurrence matrix for associating and clustering term and/or documents. Symmetric and nonsymmetric proximity measures are considered and theoretical properties of them are investigated. Normalization and symmetrization of the measures are also considered. Illustrative examples including bibliographic citations are given.