---
title: "Duce, An Oracle-based Approach to Constructive Induction"
authors:
  - "Stephen H. Muggleton"
doi: ""
url: "https://ijcai.org/proceedings/1987/59"
pdf: "https://ijcai.org/proceedings/1987/0059.pdf"
venue: "Proceedings of the Tenth International Joint Conference on Artificial Intelligence"
year: 1987
pages: "287-291"
track: "Knowledge Acquisition"
keywords:
---

# Duce, An Oracle-based Approach to Constructive Induction

Stephen H. Muggleton

*Proceedings of the Tenth International Joint Conference on Artificial Intelligence*, pages 287-291.

## Abstract

Duce is a Machine Learning ayllem which augaella highlevel domain features to the uaer (or or«lc) on the baaia d a act of example object descriptiona. Six transformation oparaton are uaed to successively compre11 the given examplea by generaliaation and feature construction. In this papar Duce Ia illuatraled by way d ill con~truction of a simple animal taxonomy and a hierarchical parity checker. However, Duce's main achievement baa been the restructuring of a substantial expert system for deciding whether )IOIIilions within the chess endgame of King-and-Pawn-on-a7 v. King-andRook (KPa7KR) are won-for-white or not. The new conceptaauggested by Duce for the che11 expert system hierarchy were found to be meaningful by the che11 expert Ivan Dratko. An exiatlng manually created KPa7KR solution, which was the basis d a recent PhD. thesis, is compared to the ltructure interactively created by Duce.