My PhD thesis was created during 1995 in WordPerfect. Ah the golden days before Microsoft swallowed the software industry! I donated a printed copy, as required by the University of Nottingham regulations, for deposit at the British Library. However, I believe that the BL decided from 1996 onwards that it wouldn't have room to store UK doctoral theses any more, so as far as I know those 300 pages or so went into a shredder somewhere.
People do occasionally want to read it, even now, so after letting it languish for over 16 years I have finally got round to converting it from 13 WordPerfect files into a single MS/Word (then pdf) file. What an effort! You wouldn't believe the fiendishly ingenious ways MS/Word finds to mess up the layout and hide the figures. Well, if you're a Linux user, you probably would. Anyway, after much sweat, I have restored it to nearly but not quite mint condition, as of 8/12/2012, correcting a few typos in the process.
It can be downloaded in 2 parts: thesis body text including Reference List, with Appendices in a separate file. See below.
Please find my PhD thesis here.
Appendices can be found here.
Summary of Contents:
Front Matter
1. Introduction
2. Background
3. Text Classification by Cross-Codebook Compression
4. Text Classification by Diagnostic Digrams
5. Instance-Based Text Classification
6. Text Classification by Rule Induction
7. Extending Textual Fragments to their Natural Lengths
8. An Evolutionary Approach to Text Classification
9. The Federalist Revisited (Again): a Case Study
10. Concluding Discussion
Reference List
Appendices