TREAT

Book description

TREAT: A New and Efficient Match Algorithm for AI Production Systems describes the architecture and software systems embodying the DADO machine, a parallel tree-structured computer designed to provide significant performance improvements over serial computers of comparable hardware complexity in the execution of large expert systems implemented in production system form.

This book focuses on TREAT as a match algorithm for executing production systems that is presented and comparatively analyzed with the RETE match algorithm. TREAT, originally designed specifically for the DADO machine architecture, handles efficiently both temporally redundant and non-temporally redundant production system programs.

This publication is suitable for developers and specialists interested in match algorithms for AI production systems.

Table of contents

  1. Front Cover
  2. Treat: A New and Efficient Match Algorithm for AI Production Systems
  3. Copyright Page
  4. Table of Contents
  5. List of Figures
  6. Acknowledgements
  7. Abstract
  8. Chapter 1. Introduction
    1. 1.1 The Problem
    2. 1.2 Outline of the Book
    3. 1.3 Digest of Conclusions
  9. Chapter 2. Background
    1. 2.1 Production Systems
    2. 2.2 Characteristics of Computer Architectures (1/2)
    3. 2.2 Characteristics of Computer Architectures (2/2)
  10. Chapter 3. The Design of Production System Algorithms
    1. 3.1 The Correspondence Between Production Systems and Relational Databases
    2. 3.2 Three Issues in Parallel Production System Algorithms
    3. 3.3 The Development of Matching algorithms
    4. 3.4 The Details of the RETE Match (1/2)
    5. 3.4 The Details of the RETE Match (2/2)
  11. Chapter 4. TREAT: A New Match Algorithm
    1. 4.1 The Motivation for TREAT
    2. 4.2 Conflict Set Support
    3. 4.3 The TREAT Algorithm (1/2)
    4. 4.3 The TREAT Algorithm (2/2)
    5. 4.4 Dynamic Ordering of the Joins
    6. 4.5 Why TREAT is Expected to Perform Well
    7. 4.6 Comparative Performance of TREAT and RETE on a Sequential Machine
    8. 4.7 Conclusion
  12. Chapter 5. The DADO Machine
    1. 5.1 The System Architecture
    2. 5.2 The DADO Prototypes
    3. 5.3 Evaluation of Alternative Designs (1/3)
    4. 5.3 Evaluation of Alternative Designs (2/3)
    5. 5.3 Evaluation of Alternative Designs (3/3)
    6. 5.4 Programming DADO
    7. 5.5 The EPROM Resident Kernel
    8. 5.6 PPL/M (1/2)
    9. 5.6 PPL/M (2/2)
  13. Chapter 6. Other Parallel AI Machines
    1. 6.1 Semantic Net-Based Machines
    2. 6.2 Logic Based Machines
    3. 6.3 Parallel Lisps (1/2)
    4. 6.3 Parallel Lisps (2/2)
    5. 6.4 Discussion
  14. Chapter 7. The Parallel Implementation of OPS5 on DADO
    1. 7.1 Related Parallel Production System Efforts
    2. 7.2 Limitations of OPS5 as a Production System Model
    3. 7.3 The Parallel Implementation of TREAT
    4. 7.4 Expected Performance
    5. 7.5 Parallelism and the Three Phases of the Production System Cycle (1/2)
    6. 7.5 Parallelism and the Three Phases of the Production System Cycle (2/2)
    7. 7.6 Conclusions
  15. Chapter 8. Conclusions and Future Research
    1. 8.1 Ongoing and Future Research
    2. 8.2 Conclusions
  16. Appendix A: LISP Code for Seed Ordered TREAT (1/2)
  17. Appendix A: LISP Code for Seed Ordered TREAT (2/2)
  18. Appendix B: Implementation Outline for PM-level TREAT in PPSL
  19. References (1/2)
  20. References (2/2)

Product information

  • Title: TREAT
  • Author(s): Daniel P. Miranker
  • Release date: July 2014
  • Publisher(s): Morgan Kaufmann
  • ISBN: 9781483258898