The characterisation of text corpora using an input/output schema for citations

Abstract

Univariate measures of concentration (or dispersion) can be applied to the description of the citation patterns within a text corpus, and also the citation links between that corpus and an alternative (possibly contextual) literature. To assist in this, a simple data-flow schema introduced byLano to assist with the design of software can be used to provide an appropriate data-definitional tool. The schema, as applied here, comprises: (1) a matrix of cells containing 0 or 1 values (in its non-diagonal cells) representing within-corpus citations, with the diagonal cells representing the corpus documents; and (2) two associated vectors of cells which record the total numbers of citations that link the corpus documents with an external-to-corpus literature. An initial data-exploration based on a application of this schema to a trial document corpus is reported. On this basis, several provisional conjectures are put forward to attract further research on data of this type. These conjectures include: (1) Concentration amongst citationsto corpus items from within a young corpus is less than it is amongst citationsby corpus items to that corpus; (2) A young literature corpus imports significantly more information from its external world than it exports to it; and (3) Information transfer from an into contextual literature dominates within-corpus information transfer. The author emphasises that these are conjectures at this stage, not hypotheses.