Dissertation Information for Michael Shepherd
- Michael Shepherd
- (Alias) Michael Alan Shepherd
- Library and Information Science
- University of Western Ontario (Canada) (1978)
- Jean Mary Tague-Sutcliffe
- William J. Cameron
MPACT Status: Complete - Except Indecipherables
Title: The Colon Classification as a Basis for Automated Full Text Retrieval
Abstract: The fundamental hypothesis tested in this research is that the Colon classification is a suitable foundation for the automated analysis, representation, and retrieval of primary information from the full text of documents. Primary information is that information embodied in the text of a document, as opposed to secondary information which is generally in such forms as: an abstract, a table of contents, or an index.
The Colon Classification is a fully faceted classification scheme illustrating the postulates and principles of the General Theory of Library Classifications as developed in India. The classification has several features which suggest that it might be a suitable foundation for a passage retrieval system. It has in-depth classification schedules for specific subject areas, its schedules are of a hierarchical structure which allows access to subordinate entries, and it provides a set of postulates and principles for the consistent arrangement of facets of a subject.
In order to test the suitability of the Colon Classification, a set of retrieval experiments were conducted. Full text data bases were created in the subject areas of programming, languages and food technology. Ten documents for each data base were selected, randomly, from the literature. In in-depth schedules for these areas were available in the literature. Queries were solicited from five subject specialists in each area, for a total of 23 queries in programming languages and 29 in food technology. An automated fell text indexing system, along with four automated passage retrieval systems, was created to test, explicitly, the various features of the Colon Classification. Two Boolean-based systems and one simple word occurrence system were created n order to compare the retrieval results against types of systems which are in more common use.
The systemsí retrieval performances were measured using recall and precision and the mean expected search length reduction factors. These retrieval results indicate that there was no significant difference between the retrieval performances of the Colon Classification systems and the performance of the simple word occurrence system, and that the Colon Classification performed with higher recall but lower precision than the Boolean systems. It must be pointed out that, within the framework of this set of experiments, the recall levels of the Boolean systems were so low that, despite the high precision levels, the overall effectiveness of these systems would seem to be lower than that of the Colon Classification systems. This was borne out by the mean expected search length reduction factors.
Overall, it was concluded that, as the Colon Classification systems did not perform significantly better than the other systems despite certain methodological biases in their favour and were considerably more expensive to operate than the simple word occurrence system, a great deal more research is required before the colon Classification can be considered as a foundation for the cost-beneficial retrieval of primary information from the full text of documents. As an area for further research and development, however, it would appear to be fertile indeed.
MPACT Scores for Michael Shepherd
A = 0
C = 1
A+C = 1
T = 0
G = 0
W = 0
TD = 0
TA = 0
calculated 2008-01-31 06:26:51
Advisors and Advisees Graph