Copyright (C) 1997  Katsumi Inoue 

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%%% Readme-E                 %%%
%%%         by Katsumi Inoue %%%
%%%         (1997.5.9)       %%%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

Development of Action Language Systems
--- Action-MGTP (Prolog version) ---

by
  Katsumi Inoue
  Department of Electrical and Electronics Engineering 
  Kobe University 
and 
  Tomokazu Yamamoto
  Department of Information and Computer Sciences 
  Toyohashi University of Technology 

Abstract: 
  Reasoning about action domains in which environment and
  knowledge change dynamically by executing action is a classical
  problem in AI, and recently it has become a hot subject for agent and
  robot planning.  For this problem, it is known that nonmonotonic
  reasoning and abduction are useful technique to cope with
  the frame problem and prediction problem.  However, there has been few
  previous research that realized reasoning about action on computers. 

  As a method of reasoning about action domains, there are high-level 
  action languages.  They are simple and declarative languages,
  and can represent changes of states by executing action.  We here
  propose a new action language, which permits to represent 
  nondeterministic actions and static constraints by extending
  the language ${\cal A}$ proposed by Gelfond and Lifschitz.  We also
  consider a translation of descriptions in our action language into
  extended disjunctive logic programs.  

  Using these techniques, we implemented the action language system, 
  called Action-MGTP, on top of our abductive logic programming (ALP) 
  system.  The ALP system is a reasoning system for logic programs 
  containing negation as failure (NAF), classical negation, disjunction, 
  and abducible literals.  It is based on bottom-up, incremental, 
  backtrack-free computation of the minimal models for disjunctive 
  programs using a model generation theorem prover (MGTP).  

  Furthermore, to make the action system more efficient, we improve 
  the efficiency of our ALP system by introducing negative clauses to
  compute negation as failure.  The experimental results with some
  action domains show that our action system has a good performance.

  Consequently, we realize a query answering system to reason about
  action in dynamic domains automatically.  


%%% This directory contains the following files: 

1. Readme-E         This file
2. Readme-J         Japanese version of this file
3. advertise.txt    Brief introduction of Action-MGTP (in Japanese)
4. copyright.txt    Copyright owner of Action-MGTP (in Japanese)
5. function.txt     Main features of Action-MGTP (in Japanese) 
6. image.txt        Image of users of Action-MGTP (in Japanese)
7. manual  <dir>    Directory containing Technical Report and Manual 
8. publication.txt  Papers describing Action-MGTP (in Japanese)
9. source  <dir>    Directory containing software and examples


%%%%%%%%%%%%
%%% NOTE %%%
%%%%%%%%%%%%

There has been no English manual for Action-MGTP.  
If you are interested in using Action-MGTP, please 
contact the copyright owner: 

  Katsumi Inoue
  Department of Electrical and Electronics Engineering 
  Kobe University 
  Rokkodai-cho, Nada-ku, Kobe  657,  Japan

  Email: inoue@eedept.kobe-u.ac.jp
  Phone: +81-78-803-1079
  Fax:   +81-78-881-3193

-------------------------------------------------------------------
9 May 97 - Katsumi Inoue, inoue@eedept.kobe-u.ac.jp
