Case-based framework for collaborative semantic search

• US 8,386,485 B2
• Filed: 07/31/2009
• Issued: 02/26/2013
• Est. Priority Date: 07/31/2008
• Status: Active Grant

• Current Assignee: George Mason Intellectual Properties, Inc.
• Original Assignee: George Mason Intellectual Properties, Inc.
• Inventors: Kerschberg, Larry, Kim, Wooju, Jeong, Hanjo, Song, Yong Uk
• Primary Examiner(s): NGUYEN, MERILYN P
• Application Number:
  US12/533,226
• Publication Number:
  US 20110029514A1
• Time in Patent Office: 1,306 Days
• Field of Search: 707/706, 707/707, 707/711, 707/713, 707/721, 707/722, 707/723, 707/732, 707/736, 707/748, 707/705, 707/718, 707/741, 707/751, 707765-768, 707/769
• US Class Current: 707/736
• CPC Class Codes: G06F 16/3326 using relevance feedback from the user, e.g. relevance feedback on documents, documents sets, document terms or passages
• CPC Class Codes: G06F 16/90324 using system suggestions

Abstract

A service-oriented system architecture includes a computer-implemented search method and computer-implemented agent system for enabling efficient information searches on, for example, on XML databases, relational databases, and files located on intranets, the Internet, or other computer network systems. Referred to as the Knowledge Sifter architecture, the architecture may comprise, in one embodiment, a community of cooperating agents. The system architecture may be employed using a variety of methodologies, such as a case-based framework, collaborative filtering, or hybrid filtering. The case-based framework may be configured to recommend query specifications and refinements based on previously-stored user-query cases. Collaborative filtering involves the architecture recommending a set of unseen items that are preferred by other users to the active user. Hybrid filtering combines collaborative filtering and content-based filtering.

Claims

1. A non-transitory physical machine readable storage medium having stored thereon instructions that that when executed by one or more processors, causes the one or more processors to perform a method comprising:

a. receiving at least one query from at least one user;

b. accepting at least one user preference from the at least one user using at least one user processor;

c. refining the at least one query using the at least one user preference using a refining processor;

d. reformulating the at least one query through semantic mediation information using at least one ontology processor;

e. retrieving at least one user query case similar to the at least one query from at least one user query case base using at least one case management processor;

f. maintaining at least one ontology-based index using the at least one case management processor;

g. storing at least one pre-compiled user query, or at least one artifact, or a combination thereof, in at least one repository;

h. decomposing the at least one query into at least one subquery using the semantic mediation information utilizing at least one query formulation processor; and

i. executing the at least one subquery in order to obtain at least one search result; and

wherein the at least one case management processor retrieves the at least one user query case similar to the at least one query from the at least one user query case base using at least one algorithm using at least one ontology-based index; and

wherein the at least one algorithm limits a number of user query cases retrieved to a predefined maximum.

2. The non-transitory physical machine readable storage medium of claim 1, wherein the method further comprises:

a. dispatching at least one subquery to at least one data source using at least one web services processor;

b. managing the at least one user preference using a preferences processor;

c. ranking at least one subquery result according to user preferences using at least one ranking processor; and

d. transmitting the at least one search result to the at least one user for displaying on at least one graphical user interface.

3. The non-transitory physical machine readable storage medium of claim 2, wherein the method further comprises:

a. receiving feedback from the at least one user based on the at least one search result displayed;

b. generating at least one user query case based on the feedback; and

c. storing the at least one user query case in the at least one repository.

4. The non-transitory physical machine readable storage medium of claim 3, wherein the method further comprises recommending at least one data item to the at least one user based on:

a. feedback from at least one prior user;

or

b. feedback from the at least one user;

or

c. any combination thereof.

5. The non-transitory physical machine readable storage medium of claim 4, wherein the at least one recommended data item is based on:

a. aggregated feedback from the at least one prior user;

or

b. aggregated feedback from the at least one user;

or

c. any combination thereof.

6. The non-transitory physical machine readable storage medium of claim 1, wherein the at least one ontology-based index is comprised of at least one of the following:

a. an ontology identifier;

b. an concept index;

or

c. a case identifier;

or

d. any combination thereof.

7. A computer-implemented search method, the method comprising:

a. receiving at least one query from at least one user;

b. accepting at least one user preference from the at least one user using at least one user processor;

c. refining the at least one query using the at least one user preference using a refining processor;

d. reformulating the at least one query through semantic mediation information using at least one ontology processor;

e. retrieving at least one user query case similar to the at least one query from at least one user query case base using at least one case management processor;

f. maintaining at least one ontology-based index using the at least one case management processor;

g. storing at least one pre-compiled user query, or at least one artifact, or a combination thereof, in at least one repository;

h. decomposing the at least one query into at least one subquery using the semantic mediation information utilizing at least one query formulation processor; and

i. executing the at least one subquery in order to obtain at least one search result; and

wherein the at least one case management processor retrieves the at least one user query case similar to the at least one query from the at least one user query case base using at least one algorithm using at least one ontology-based index; and

wherein the at least one algorithm limits a number of user query cases retrieved to a predefined maximum.

8. The computer-implemented search method of claim 7, the method further comprising:

a. dispatching at least one subquery to at least one data source using at least one web services processor;

b. managing the at least one user preference using a preferences processor;

c. ranking at least one subquery result according to at least one user preference using a ranking processor; and

d. transmitting the at least one search result to the at least one user for display on at least one graphical user interface.

9. The computer-implemented search method of claim 8, wherein the at least one data source is selected from at least one of the following:

a. Yahoo! Images;

b. TerraServer;

c. the Semantic Web;

or

d. the World Wide Web;

e. or any combination thereof.

10. The computer-implemented search method of claim 8, wherein the preferences processor stores user attributes comprising at least one of the following:

a. a weight associated with the at least one data source based on a predefined weighting scale;

b. a weight associated with at least one term comprising the query;

or

c. any combination thereof.

11. The computer-implemented search method of claim 7, the method further comprising:

a. receiving feedback from the at least one user based on the at least one search result displayed;

b. generating at least one user query case based on the feedback; and

c. storing the at least one user query case in at least one repository.

12. The computer-implemented search method of claim 11, wherein the feedback received from the user includes a rating of at least one of the following:

a. highly relevant;

b. relevant;

c. highly not relevant;

or

d. unclear;

or

e. any combination thereof.

13. The computer-implemented search method of claim 7, the method further comprising:

a. accessing at least one domain model database using the ontology processor;

b. consulting at least one authoritative name service using the ontology processor; and

c. providing at least one query refinement to the query formulation processor using the at least one authoritative name service.

14. The computer-implemented search method of claim 13, wherein the at least one authoritative name service is selected from at least one of the following:

a. WordNet;

b. GNIS;

or

c. GNS;

or

d. any combination thereof.

15. The computer-implemented search method of claim 13, wherein the domain model is specified in web ontology language.