Process for semi-automatic maintenance of a knowledge base using tagged examples

• US 7,401,063 B2
• Filed: 12/16/2002
• Issued: 07/15/2008
• Est. Priority Date: 12/16/2002
• Status: Expired due to Fees

• Current Assignee: ABB Schweiz AG (ABB Limited)
• Original Assignee: General Electric Company
• Inventors: Cuddihy, Paul Edward, Azzaro, Steven Hector, Yu, Lijie, Cleary, Daniel Joseph, Donoghue, Jeremiah Francis, Johnson, Timothy Lee
• Primary Examiner(s): STARKS, WILBERT L
• Application Number:
  US10/318,682
• Publication Number:
  US 20040117335A1
• Time in Patent Office: 2,038 Days
• Field of Search: 706/50, 707/100, 707/102, 707/2, 715/512, 715/517, 704/9
• US Class Current: 706/50
• CPC Class Codes: G06N 5/025 Extracting rules from data

Abstract

The present invention provides a computer implemented method for maintaining a knowledge base. The method taking as input, text examples that are tagged with a structural markup language; and maps knowledge nodes in the knowledge base with the tagged examples to determine: (1) the knowledge nodes that best match the tagged examples, and (2) the knowledge nodes that are best connected to the tagged examples. The results are displayed to the user, who verifies that the tagged examples match the selected knowledge nodes. Based on the user response, indices of the knowledge nodes are updated. The method may further include determining discrepancies between the existing knowledge base and the tagged examples, and further displaying the discrepancies to the user. The user can update the missing information in the knowledge base to remove the discrepancies in the existing knowledge base.

Claims

1. A computer implemented method for maintaining and updating a knowledge base, the knowledge base arranged in network structures and consisting of knowledge nodes, each of the knowledge nodes being further indexed using text matching methods, with examples that are tagged with a markup language, the method comprising:

(a) receiving the tagged examples into memory of a computer from a user;

(b) mapping the tagged examples to the knowledge nodes, including determining the knowledge nodes that are best matched to the tagged examples, and determining the knowledge nodes that are best connected to the tagged examples;

(c) updating the knowledge base'"'"'s indices;

(d) determining the discrepancies between the knowledge nodes and the tagged examples; and

(e) storing the discrepancies in the computer memory and displaying the discrepancies.

2. The method as recited in claim 1 wherein determining the knowledge nodes that are best matched to the tagged examples comprises the steps of:

(a) determining the roots of words in the tagged examples;

(b) calculating frequency of occurrence of the roots in the tagged examples; and

(c) calculating the frequency of occurrence of roots to the frequency of words in the knowledge base.

3. The method as recited in claim 1 wherein determining the knowledge nodes that are best connected to the tagged examples comprises the steps of:

(a) determining the network structure with the maximum number of matching knowledge nodes; and

(b) displaying the best connected network structures in decreasing order of relevance.

4. The method as recited in claim 3 wherein the step of displaying the best-connected network structures in decreasing order of relevance, displaying the matched and mismatched portions of the network structures, if there are multiple matches.

5. A computer implemented method for maintaining and updating a knowledge base, the knowledge base arranged in network structures and consisting of knowledge nodes, each of the knowledge nodes being further indexed using text matching methods, with examples that are tagged with a markup language, the method comprising:

(a) receiving the tagged examples into memory of a computer from a user;

(b) mapping the tagged examples to the knowledge nodes;

wherein the step of mapping the tagged examples further comprises the steps of;

i) determining the knowledge nodes that are best matched to the tagged examples;

wherein the step of determining the knowledge nodes that are best matched to the tagged examples comprises the steps of;

1. determining the roots of words in the tagged examples;

2. calculating frequency of occurrence of the roots in the tagged examples; and

3. calculating the frequency of occurrence of roots to the frequency of words in the knowledge base; and

ii) determining the knowledge nodes that are best connected to the tagged examples;

wherein the step of determining the knowledge nodes that are best connected to the tagged examples comprises the steps of;

1. determining the network structure with the maximum number of matching knowledge nodes; and

2. displaying the best connected network structures in decreasing order of relevance;

wherein the step of displaying the best-connected network structures in decreasing order of relevance, comprises the step of displaying the matched and mismatched portions of the network structures, if there are multiple matches;

(c) updating the knowledge base'"'"'s indices; and

(d) determining the discrepancies between the knowledge nodes and the tagged examples, storing the discrepancies in the computer memory, and displaying the discrepancies.

6. A computer system for maintaining and updating a knowledge base, the knowledge base arranged in network structures and consisting of knowledge nodes, each of the knowledge nodes being indexed using text matching methods, with examples that are tagged with a markup language, comprising:

(a) a input portion that receives the tagged examples into memory of a computer from a user;

(b) a mapping portion for comparing the tagged examples with the knowledge nodes, including determining the knowledge nodes that are best matched to the tagged examples, and determining the knowledge nodes that are best connected to the tagged examples;

(c) a portion for determining discrepancies between the knowledge nodes and the tagged examples;

(d) a modifying portion for updating the knowledge base'"'"'s indices; and

(e) a portion for storing the discrepancies in the computer memory and displaying the discrepancies.

7. A computer program product for use with a computer, the computer program product comprising a computer usable medium having a computer readable program code embodied therein for maintaining and updating a knowledge base, the knowledge base stored on a computer readable medium and arranged in network structures and consisting of knowledge nodes, each of the knowledge nodes being further indexed using text matching methods, with examples that are tagged with a markup language, the computer program code performing the steps of:

(a) receiving the tagged examples into memory of a computer from a user;

(b) mapping the tagged examples to the knowledge nodes, including determining the knowledge nodes that are best matched to the tagged examples, and determining the knowledge nodes that are best connected to the tagged examples;

(c) updating the knowledge base'"'"'s indices;

(d) determining the discrepancies between the knowledge nodes and the tagged examples; and

(e) storing the discrepancies in the computer memory and displaying the discrepancies.

8. The computer program product as recited in claim 7, wherein determining the knowledge nodes that are best matched to the tagged examples comprises the steps of:

(a) determining the roots of words in the tagged examples;

(b) calculating frequency of occurrence of the roots in the tagged examples; and

(c) calculating the frequency of occurrence of roots to the frequency of words in the knowledge base.

9. The computer program product as recited in claim 7, wherein determining the knowledge nodes that are best connected to the tagged examples comprises the steps of:

(a) determining the network structure with the maximum number of matching knowledge nodes; and

(b) displaying the best connected network structures in decreasing order of relevance.

10. The computer program product as recited in claim 9, wherein the step of displaying the best-connected network structures in decreasing order of relevance, comprises the step of displaying the matched and mismatched portions of the network structures, if there are multiple matches.

11. A computer program product for use with a computer, the computer program product comprising a computer usable medium having a computer readable program code embodied therein maintaining and updating a knowledge base, the knowledge base stored on a computer readable medium and arranged in network structures and consisting of knowledge nodes, each of the knowledge nodes being further indexed using text matching methods, with examples that are tagged with a markup language, the computer program code performing the steps of:

(a) receiving the tagged examples into memory of a computer from a user;

(b) mapping the tagged examples to the knowledge nodes;

wherein the step of mapping the tagged examples further comprises the steps of;

i) determining the knowledge nodes that are best matched to the tagged examples;

wherein the step of determining the knowledge nodes that are best matched to the tagged examples comprises the steps of;

1. determining the roots of words in the tagged examples;

2. calculating frequency of occurrence of the roots in the tagged examples; and

3. calculating of the frequency of occurrence of roots to the frequency of words in the knowledge base; and

ii) determining the knowledge nodes that are best connected to the tagged examples;

wherein the step of determining the knowledge nodes that are best connected to the tagged examples comprises the steps of;

1. determining the network structure with the maximum number of matching knowledge nodes; and

2. displaying the best connected network structures in decreasing order of relevance;

wherein the step of displaying the best-connected network structures in decreasing order of relevance, comprises the step of displaying the matched and mismatched portions of the network structures, if there are multiple matches;

(c) updating the knowledge base'"'"'s indices; and

(d) determining the discrepancies between the knowledge nodes and the tagged examples, storing the discrepancies in the computer memory, and displaying the discrepancies.