Method and Apparatus for Processing Service Requests

• US 20170221015A1
• Filed: 01/29/2016
• Published: 08/03/2017
• Est. Priority Date: 01/29/2016
• Status: Active Grant

• Current Assignee: The Boeing Co.
• Original Assignee: The Boeing Co.
• Inventors: June, Philip Gabriel, Yu, Bryan Quolent, Nguyen, Thai Thanh, Chan, Candice Sik Suen, Danak, Amir, Jiang, Ruiwei
• Granted Patent: US 10,366,371 B2
• Time in Patent Office: Days
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06Q 10/06315 Needs-based resource requirements planning or analysis
• CPC Class Codes: G06Q 10/20 Administration of product repair or maintenance
• CPC Class Codes: G06Q 30/016 After-sales

Abstract

A method and apparatus for processing service requests. A computer system creates a preliminary compendium comprising building blocks. A building block is a word entity extracted from a text source in a plurality of text sources that are related to a domain of interest. At least a portion of the text sources comprises past service requests. The computer system filters the building blocks in the preliminary compendium based on relevance with respect to the domain of interest to create similarity indicators that form a refined compendium. The computer system analyzes text within a new service request using the past service requests and the refined compendium to generate a ranking of past service requests that ranks the past service requests based on similarity to the new service request. The ranking of past service requests enables the computer system to identify a set of possible resolutions for the new service request.

Claims

1. A method for improving a processing speed of processing service requests, the method comprising:

creating, by a computer system, a preliminary compendium comprising a plurality of building blocks, wherein a building block in the plurality of building blocks is a word entity extracted from a text source in a plurality of text sources that are related to a domain of interest and wherein at least a portion of the plurality of text sources comprises a plurality of past service requests;

filtering, by the computer system, the plurality of building blocks in the preliminary compendium based on relevance with respect to the domain of interest to create a plurality of similarity indicators that form a refined compendium; and

analyzing, by the computer system, text within a new service request using the plurality of past service requests and the refined compendium to generate a ranking of past service requests that ranks the plurality of past service requests based on similarity to the new service request,

wherein the ranking of past service requests enables the computer system to identify a set of possible resolutions for the new service request.

2. The method of claim 1, wherein creating, by the computer system, the preliminary compendium comprises:

selecting a text source that is unprocessed from the plurality of text sources for processing, wherein the text source is either a past service request in the plurality of past service requests or an article; and

determining whether at least one anchor from a plurality of selected anchors is present in the text source.

3. The method of claim 2, wherein creating, by the computer system, the preliminary compendium further comprises:

identifying a neighborhood of text around each anchor identified in the text source in response to a determination that at least one anchor is present in the text source; and

adding each word entity found in the neighborhood of text as a building block to the preliminary compendium.

4. The method of claim 3, wherein creating, by the computer system, the preliminary compendium further comprises:

repeating the selecting step, the determining step, the identifying step, and the adding step for each text source in the plurality of text sources to create the preliminary compendium.

5. The method of claim 1 further comprising:

adding the new service request to the plurality of past service requests after the new service request has been resolved.

6. The method of claim 1, wherein filtering, by the computer system, the plurality of building blocks comprises:

computing an entropy for each building block in the plurality of building blocks in the preliminary compendium to form a plurality of entropies.

7. The method of claim 6, wherein filtering, by the computer system, the plurality of building blocks further comprises:

adding a selected building block to the refined compendium as a similarity indicator if the entropy computed for the selected building block is below a selected threshold.

8. The method of claim 1 further comprising:

performing, by the computer system, association rule mining using the plurality of similarity indicators in the refined compendium to form a plurality of combination indicators.

9. The method of claim 8, wherein performing, by the computer system, the association rule mining comprises:

mapping the plurality of similarity indicators to the plurality of text sources; and

identifying the plurality of combination indicators based on mapping the plurality of similarity indicators to the plurality of text sources and a set of association rule learning algorithms, wherein a combination indicator in the plurality of combination indicators is formed by a combination of at least two similarity indicators from the plurality of similarity indicators.

10. The method of claim 1 further comprising:

identifying, by the computer system, a set of possible resolutions to the new service request based on a highest ranked past service request in the ranking of past service requests.

11. The method of claim 1, wherein analyzing, by the computer system, the text within the new service request comprises:

identifying a set of indicators that are present in the new service request using the refined compendium, wherein an indicator in the set of indicators is one of a similarity indicator and a combination indicator; and

generating a similarity score for each pairing of the new service request and a corresponding past service request in the plurality of past service requests based on the set of indicators identified within the new service request.

12. The method of claim 11, wherein analyzing, by the computer system, the text within the new service request further comprises:

ordering the plurality of past service requests by the similarity score to form the ranking of past service requests.

13. An apparatus comprising:

a trainer implemented in a computer system that creates a preliminary compendium comprising a plurality of building blocks, wherein a building block in the plurality of building blocks is a word entity extracted from a text source in a plurality of text sources that is related to a domain of interest and wherein at least a portion of the plurality of text sources comprises a plurality of past service requests and wherein the trainer filters the plurality of building blocks in the preliminary compendium based on relevance with respect to the domain of interest to create a plurality of similarity indicators that form a refined compendium; and

an analyzer implemented in the computer system that analyzes text within a new service request using the plurality of past service requests and the refined compendium to generate a ranking of past service requests that is ranked based on similarity to the new service request,

wherein the ranking of past service requests enables the computer system to identify a set of possible resolutions for the new service request.

14. The apparatus of claim 13 further comprising:

a data structure that stores the plurality of text sources.

15. The apparatus of claim 13, wherein the new service request is received from a customer system in communication with the analyzer.

16. The apparatus of claim 13, wherein the domain of interest is aircraft maintenance.

17. The apparatus of claim 13, wherein the new service request is received from an airline customer.

18. The apparatus of claim 13, wherein the new service request is received by an aircraft manufacturer.

19. The apparatus of claim 13 further comprising:

the preliminary compendium storing the plurality of building blocks; and

the refined compendium storing the plurality of similarity indicators.

20. The apparatus of claim 13, wherein the ranking of past service requests includes a similarity score for each past service request in the plurality of past service requests in which the similarity score for a particular past service request in the plurality of past service requests measures a similarity between the new service request and the particular past service request based on a set of indicators identified in the new service request.

21. The apparatus of claim 13, wherein the trainer and the analyzer form a request processing system.

22. The apparatus of claim 13 further comprising:

a graphical user interface displayed on a display system for displaying the ranking of past service requests.

23. A request processing system implemented in a computer system, wherein the request processing system comprises:

a data structure storing a plurality of text sources related to aircraft maintenance, wherein the plurality of text sources includes a plurality of past service requests received from an airline customer;

a trainer that creates a preliminary compendium comprising a plurality of building blocks and filters the plurality of building blocks in the preliminary compendium based on relevance with respect to the aircraft maintenance to create a plurality of similarity indicators that form a refined compendium,

wherein a building block in the plurality of building blocks is a word entity extracted from a text source in the plurality of text sources; and

an analyzer that analyzes text within a new service request received over at least one communications link from the airline customer using the plurality of past service requests and the refined compendium to generate a ranking of past service requests that is ranked based on similarity to the new service request,

wherein the ranking of past service requests enables the computer system to identify a set of possible resolutions for the new service request.