Related Work

Abstract

While automated theory formation in mathematics has not been the most researched topic in Artificial Intelligence, there is some previous work against which we can evaluate our contribution. We aim here to compare and contrast HR with previous programs in order to put our work in context. We compare HR with mathematical theory formation programs, namely AM [Lenat 82], GT [Epstein 91], IL [Sims 90] and the system from Bagai et al [Bagai et al. 93]. We also compare HR with the Graffiti program [Fajtlowicz 88] which was developed to perform discovery tasks in graph theory and the Progol machine learning program [Muggleton 95]. An overview of each of these programs has been given in Chapter 2. In this chapter, we give a brief recap of what each program did and then provide more detail in order to compare it with HR. We provide a qualitative comparison based on how the program operated and how HR operates, and where possible, a quantitative comparison based on the reported output from the program.