Browse Prior Art Database

GENERAL DECISION TREE GENERATOR AND NAVIGATOR FOR A PRODUCTION RULE SYSTEM

IP.com Disclosure Number: IPCOM000014143D
Original Publication Date: 2001-Jul-01
Included in the Prior Art Database: 2003-Jun-19
Document File: 2 page(s) / 54K

Publishing Venue

IBM

Abstract

A decision tree is a hierarchical arrangement of decision points structured to reach a solution to a problem. You can use decision trees to implement a help desk or diagnostic routine that recommends actions based on answers to questions or the existence of given conditions. Novel Advance over the Prior Art: The heart of the invention is a set of methods for providing features for a decision tree generator and navigator using a production rule system. These features are used in a sample application called DTREE for the Integrated Reasoning Shell (TIRS) Version 1 Release 3 consisting of a sample knowledge base, generation utility, and set of data files. The DTREE sample provides a foundation for several powerful decision tree features: A text format for specifying a decision tree A method and format for specifying an arbitrary number of decision trees in a fixed number of text files, and a method of permitting the coexistence of multiple decision trees using data structure instances (frames) without naming conflicts A method of navigating a decision tree with a small number of rules A method of supporting answers at a node in any supported data type A method of transferring control between decision trees A method of connecting nodes in a decision tree without branching (using limbs) A method of specifying and implementing (single-condition) demons A method of generating equivalent decision tree formats (using the DTREED.CMD utility).

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 53% of the total text.

Page 1 of 2

  GENERAL DECISION TREE GENERATOR AND NAVIGATOR FOR A PRODUCTION RULE SYSTEM

A decision tree is a hierarchical arrangement of decision points structured to reach a solution to a problem. You can use decision trees to implement a help desk or diagnostic routine that recommends actions based on answers to questions or the existence of given conditions.

Novel Advance over the Prior Art: The heart of the invention is a set of methods for providing features for a decision tree generator and navigator using a production rule system. These features are used in a sample application called DTREE for the Integrated Reasoning Shell (TIRS) Version 1 Release 3 consisting of a sample knowledge base, generation utility, and set of data files.

The DTREE sample provides a foundation for several powerful decision tree features:

A text format for specifying a decision tree A method and format for specifying an arbitrary number of decision trees in a fixed number of text files, and a method of permitting the coexistence of multiple decision trees using data structure instances (frames) without naming conflicts A method of navigating a decision tree with a small number of rules A method of supporting answers at a node in any supported data type A method of transferring control between decision trees A method of connecting nodes in a decision tree without branching (using limbs) A method of specifying and implementing (single-condition) demons A method of generating equivalent decision tree formats (using the DTREED.CMD utility).

This invention solves the problem of specifying a decision tree using a text editor, representing a decision tree in machine readable format, and navigating it using a production rule system in response to input from the user or other interface.

The DTREE sample knowledge base suggests a set of generic rules and frame types that let you create a decision tree of any content and run it by tracing through the decision tree while interacting with a user or other input sources.

In the DTREE sample, a decision tree is a hierarchical arrangement of nodes, branches, and limbs. Decision points are represented with nodes. A decision tree starts at one node which is usually at the beginning of the decision tree. The DTREE sample considers that starting node the "current node." As the application moves through decision points, the current node changes according to the answers to the questions. A node specifies on of four things:

A statement
1.

A question
2.


3.


4.

A decision tree for a manufacturing process might include decision points in a structure that looks like this:

Transfers: DTREE provides a method of transferring control between decision trees. You can use a transfer for linking and combining decision trees, each of which represent a part of a total problem. Each decision tree (with its transitions) can be l...