Artificial Intelligence Research Group

Some Past and Current Projects
Under Development

2. Automated Text Categorising
Investigator(s): Dr. Victoria Uren, Professor Max Bramer and Professor Tom Addis

Automatic text categorizing is where a computer program will scan a document and determine from its content what category or subject area it should be identified with. No system is fool proof but it is thought that if several systems are combined then improved allocation would result. This work evolves a theory of the combination of text categorizors and explores the correctness of its predictions. Exploratory research using voting systems have been explored in the light of a probabilistic model originally developed for safety critical systems. From this a formal theory has evolved which can predict the effectiveness of such systems.

phd thesis