Contextual Hierarchies in Classification Schemes

Abstract

This paper is concerned with the encoding of contextual hierarchies. In particular, such hierarchies make it possible to create a single, complete classified display of very large thesauri. This classification may use the same descriptor with different views, as evidenced by the same descriptor as more than one node in the classification, where the nodes have different sets of children. This son of knowledge organization requires special computer representation techniques using contextual indicators for encoding the parent-child relationship. One solution" is to avoid having a unified classification in favor of many hierarchical families as used by the INSPEC® and ERIC® thesauri. However, this author considers it a particular strength to have a unified classification, of which the Medical Subject Headings (MeSH®) tree structures is a primary example. This paper describes the traditional method of using tree numbers as contextual indicators. We then propose a new experimental method of semantic labels, developed for the MedIndEx™ prototype, as possibly having certain advantages over tree numbers. We conclude with the hope that this workshop will provide feedback regarding the significance of the problem and substance of our proposal.