---
title: "Generalizing Problem Reduction: A Logical Analysis"
authors:
  - "Drew V. McDermott"
doi: ""
url: "https://ijcai.org/proceedings/1983/71"
pdf: "https://ijcai.org/proceedings/1983/0071.pdf"
venue: "Proceedings of the Eighth International Joint Conference on Artificial Intelligence"
year: 1983
pages: "302-307"
track: "Knowledge Representation"
keywords:
---

# Generalizing Problem Reduction: A Logical Analysis

Drew V. McDermott

*Proceedings of the Eighth International Joint Conference on Artificial Intelligence*, pages 302-307.

## Abstract

Problem reduction is the name given to the problemsolving paradigm in which the problem solver manages a network of "tasks" representing its intentions, repeatedly reducing tasks to subtasks and coordinating their execution. This idea needs a lot of generalization for it to be able to ha ndle a realistic range of problems. Even after the model of time is made more realistic (to handle continuity and bra nching), issues remain regarding what it means to have a ta sk or a subtask, how a task can succeed or fail, whether a t,ask is feasib le. A profitable way to study these issues is to attempt to add axioms about tasks to a first-order temporal logic . The result sheds light on what sorts of generalizations of task networks are needed.