City Research Online

Abstract

In this thesis a graph theoretic model of thematic structure in the content of written documents has been examined.

The graph model is derived from the documents using a form of Content Analysis to define the major themes contained in the documents and a relation between those themes. The themes are identified with the vertices of a graph and the relation defines the edges.

The resulting graphs can be studied in terms of their invariants to provide quantitative measures of thematic structure. These measures are used to describe and compare the structure of ideas in groups of documents.

The model has been tested empirically using a substantial body of documents of the Scottish National Party over forty years and a smaller number of Scottish Conservative Party documents.

In this analysis the model proved to be direct and substantively plausible. It was found to be immediately useful in generating a new kind of structural hypothesis and in forcing clarification of concepts such as 'centrality' and ’complexity'. Individual measures, such as eccentricity were sensitive, others, such as rank degree, more robust. The usefulness of the measures was limited by a lack of substantive theory.

The assignment of undated documents was attempted. Longer documents could be accurately placed in the correct period, shorter ones were difficult to assign.

Extensions to the model towards a multi-level, full relational structure have been investigated. These would not be practicable without further computerisation. An outline architecture for a computerised model has been proposed.

The theoretical implications of a structural model for a system of ideas has been examined and the relation of this work to similar work on manifest idea-systems in cognate disciplines has been noted. The model is likely to be most useful in the empirical study of such systems.

Publication Type:	Thesis (Doctoral)
Subjects:	T Technology > T Technology (General)
Departments:	School of Science & Technology > Computer Science School of Science & Technology > School of Science & Technology Doctoral Theses Doctoral Theses

Export

Downloads