Linguistic expressions and indexing information resources.

摘要

This dissertation is an investigation into the relation between indexing and language, specifically the ways in which certain characteristics of language can undermine effective indexing. Taking these characteristics into account when considering any indexing approach is important, but especially so when statistical analyses of language are adopted for the purposes of indexing. Indexing guidelines often advise first identifying concepts and then selecting appropriate linguistic expressions to represent those concepts. When indexing information resources, the linguistic expressions found in these resources determine how concepts are defined and represented. While it is acknowledged that many concepts are non-linguistic, most concepts, especially in the context of indexing, rely heavily on language to be expressed, conveyed, and understood. Due to the fact that the same linguistic expression can be used to convey several different concepts and that one concept can be expressed with many different linguistic expressions, interpretations of linguistic expressions are often ambiguous and vague. The flexibility of language is one of its strengths, but it also contributes to the kinds of linguistic instability that make indexing difficult with any method and particularly so with automated methods. Finally, figurative uses of language provide an example of the far end of the linguistic versatility spectrum and how this works against the goals of indexing.