An evolvable personal advisor to optimise internet search technologies

Abstract

The use of AI and the application of machine learning techniques to optimize services provided by existing internet search technologies is one way to control and manage the immense and ever-increasing volume of data published on the WWW. Users demand effective and efficient on-line information access to reduce information overload. In this paper we present a novel approach to achieve these objectives by generating information which is of a high recall quality - by reusing the output generated from major search engines and other previously developed systems; and of a high precision calibre - by generating specific user profiles after several interactions with the system. This paper discusses the design issues involved, as well as practical issues such as evolvability, profile generation, and the graphic user interface.