Knowledge acquisition in expert systems

• US 20030115170A1
• Filed: 12/27/2001
• Published: 06/19/2003
• Est. Priority Date: 12/14/2001
• Status: Active Grant

• Current Assignee: Ditto AI Ltd.
• Original Assignee: Empiricom Technologies Ltd.
• Inventors: Turner, Richard, Duke, Peter
• Granted Patent: US 6,868,422 B2
• Time in Patent Office: Days
• US Class Current: 707/1
• CPC Class Codes: G06N 5/022 Knowledge engineering; Knowledge acquisition
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

A method of knowledge acquisition for entering data into a computer database in a readily computable format, comprises eliciting a set of data statements using an analysis structure determined from an underlying mathematical base, combined with a genuine falsificationist epistemology, entering elicited data statements into a database structured as a plurality of data entries, each data entry comprising a label, a said data statement and a default value, and a root node indicator; storing a relationship between said data entries as a connectivity lattice.

Claims

1. A method of data acquisition for collecting data for entry into a computer entity, said method comprising the processes of:

compiling a set of a plurality of knowledge data statements each knowledge data statement representing an element of a knowledge domain;

creating one or a plurality of lattice data, each said lattice data comprising a plurality of nodes, each said node representing a knowledge data statement, said plurality of nodes arranged as a root node and a plurality of child nodes arranged in an hierarchical structure, leading to said root node;

said lattice data defining belief states resulting from inter relationships between said plurality of knowledge data statements;

creating a consistency matrix, said consistency matrix comprising an array of data entries representing consistency relationships between said plurality of knowledge data statements;

storing said knowledge data statements in a data storage device of said computer entity;

storing said consistency matrix in a data storage device of said computer entity; and

storing said lattice data in said data storage device of said computer entity;

wherein each said knowledge data statement is obtained as a result of a knowledge elicitation process which seeks to overturn a belief statement represented by said set of nodes.

2. The method as claimed in claim 1, comprising the processes of:

inputting said plurality of knowledge data statements into said computer entity; and

inputting said lattice data into said computer entity.

3. The method as claimed in claim 1, wherein said process of creating a said lattice data comprises:

creating an hierarchical structure network comprising a plurality of nodes and a plurality of connections between said nodes, wherein

individual said knowledge data statements comprise a root node, at least one intermediate node, and at least one leaf node of said hierarchical structure; and

said plurality of connections comprise logical connectivity relationships between said plurality of nodes.

4. The method as claimed in claim 1, wherein said step of creating a lattice data comprises:

creating an hierarchical structure network comprising a plurality of nodes and a plurality of connections between said nodes, wherein

individual said knowledge data statements comprise a root node, at least one intermediate node, and at least one leaf node of said hierarchical structure;

said plurality of connections comprise logical connectivity relationships between said plurality of nodes; and

assigning to each said node, a knowledge statement name, and a root value indicator.

5. An electronic database for storing expert data, said database having an architecture comprising:

a dictionary portion; and

a connectivity map; and

a consistency matrix;

said dictionary; and

portion comprising a plurality of data entries, each data entry comprising;

a knowledge statement;

a label identifying said data entry;

a default value condition of said knowledge statement;

said connectivity map comprising a plurality of connectivity relationships between individual said data entries of said dictionary; and

said consistency matrix comprising an array of data entries representing consistency relationships between said plurality of knowledge statements.

6. The database as claimed in claim 5, further comprising:

a data entry describing a knowledge source, comprising a source of said data statements.

7. The database as claimed in claim 5, comprising:

a data entry describing a facilitator being an elicitator of said knowledge statements.

8. The database as claimed in claim 5, further comprising:

at least one data entry describing a date of elicitation of at least one of said plurality of knowledge statements.

9. The database as claimed in claim 5, comprising:

a name data entry, describing a name of said database.

10. The database as claimed in claim 5, comprising a data entry listing a computer readable file name of said database according to a computer file system.

11. The database as claimed in claim 5, comprising:

a name data entry, describing a name of said database.

12. A knowledge base comprising:

a dictionary portion;

a connectivity lattice portion; and

a consistency matrix;

said dictionary portion comprising a plurality of data entries, each data entry comprising;

a knowledge statement;

a label identifying said data entry;

a default value condition of said knowledge statement; and

said connectivity lattice comprising a plurality of connectivity relationships between individual said data entries of said dictionary portion; and

said consistency matrix comprising an array of data representing relationships between said plurality of knowledge statements.

13. The database as claimed in claim 12, wherein said connectivity data comprises:

an hierarchical arrangement of a plurality of said labels, arranged in a lattice structure wherein individual said labels are positioned at a root node, a plurality of leaf nodes, and at a plurality of intermediate nodes of said lattice structure.

14. The database as claimed in claim 12, further comprising:

a data entry identifying a knowledge source comprising a source of said data statements.

15. The database as claimed in claim 12, comprising:

a data entry identifying a facilitator being an elicitator of said knowledge statements.

16. The database as claimed in claim 12, further comprising:

at least one data entry describing at least one date of elicitation of at least one said knowledge statement(s).

17. The database as claimed in claim 12, comprising a data entry listing a computer readable file name of said database according to a computer file system.

18. The database as claimed in claim 12, wherein said connectivity data comprises:

an hierarchical arrangement of a plurality of said labels, arranged in a lattice structure wherein individual said labels are positioned at a root node, a plurality of leaf nodes, and at a plurality of intermediate nodes of said lattice structure.

19. A method of knowledge acquisition for entering data into a computer database in a readily computable format, said method comprising:

eliciting a set of data statements using a questioning structure determined from an underlying mathematical base, combined with a genuine falsificationist epistemology;

entering elicited data statements into said database wherein said data statements are structured as a plurality of data entries, each data entry comprising a label, a said data statement, and a root value indicator;

storing a relationship between said data entries as a connectivity lattice, said connectivity lattice structured in an hierarchical lattice structure; and

storing consistency data representing a set of consistency relationships between said plurality of knowledge statements, in a matrix format.

20. A method of constructing a knowledge base comprising:

constructing a lattice comprising a plurality of nodes and links by;

exploring nodes in a first dimension firstly, looking for conditions which overturn an objective statement; and

then

going back to explore nodes in a second dimension, seeking every exception to each said nodes effect on the objective statement on the root node;

noting which nodes require further explanation;

exploring each node which requires further explanation, to derive a respective sub lattice;

wherein there is built up a multi dimension lattice structure, having as a root node said objective statement.

Having as leaf nodes, knowledge statements to which there are no exceptions.

Having as a plurality of intermediate nodes knowledge statements which singly or in conjunction with one or more said nodes are exceptions to other said knowledge statements'"'"' effects on of the value of the objective statement on said root node.

21. A method of knowledge delivery for delivering knowledge stored in a knowledge base, said knowledge base comprising:

a dictionary portion;

a connectivity lattice portion; and

a consistency matrix portion;

said dictionary portion comprising a plurality of data entries, each data entry comprising a knowledge statement, a label identifying said data entry, and a default value condition of said knowledge statement; and

said connectivity lattice comprising a plurality of connectivity relationships between said individual data entries of said dictionary portion,

said method comprising the processes of;

constructing a plurality of questions, each said question based upon a said knowledge statement;

for each said question, recording an answer; and

selecting a node from said lattice according to each said answer.

22. The method as claimed in claim 21, wherein:

individual questions are constructed, according to individual knowledge statements output from said knowledge base, in an order in which said knowledge statements were originally entered into said knowledge base.