Method and apparatus for identifying and classifying query intent

• US 8,478,780 B2
• Filed: 04/23/2010
• Issued: 07/02/2013
• Est. Priority Date: 08/14/2006
• Status: Active Grant

• Current Assignee: Oracle OTC Subsidiary LLC (Oracle Corporation)
• Original Assignee: Oracle OTC Subsidiary LLC (Oracle Corporation)
• Inventors: Cooper, Edwin Riley, Dukes, Michael Peter, Bierner, Gann Alexander, Beghelli, Filippo Ferdinando Paulo
• Primary Examiner(s): Shmatov, Alexey
• Application Number:
  US12/766,183
• Publication Number:
  US 20100205180A1
• Time in Patent Office: 1,166 Days
• Field of Search: 707/999.003, 707/771, 707/778
• US Class Current: 707/771
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/3344 using natural language analysis
• CPC Class Codes: G06F 40/30 Semantic analysis
• CPC Class Codes: G06F 40/35 Discourse or dialogue representation
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/02 Knowledge representation; Symbolic representation

Abstract

Linguistic analysis is used to identify queries that use different natural language formations to request similar information. Common intent categories are identified for the queries requesting similar information. Intent responses can then be provided that are associated with the identified intent categories. An intent management tool can be used for identifying new intent categories, identifying obsolete intent categories, or refining existing intent categories.

Claims

1. A computer-readable memory having instructions stored thereon that, when executed by a computing device, cause the computer device to perform operations comprising:

providing an ontology that defines a hierarchy of a plurality of concepts at different concept domain layers and that links natural language words to the plurality of concepts at different concept domain layers;

classifying each of a plurality of received queries with a particular intent category based on the ontology, wherein the intent category is pre-defined by a single natural language word or phrase that summarizes in a same natural language word formation a generalized information requested by the queries and wherein the queries use different natural language formations to request different types of information summarized by the single natural language word or phrase of the particular intent category;

associating an intent response formulated of natural language words or phrases with the particular intent category, wherein the intent response provides a single common applicable response for the queries;

providing an intent hierarchy of a plurality of intent categories including the particular intent category, wherein natural language phrases representing the hierarchy of intent categories at different domain layers of the intent hierarchy have different generalities of terms than natural language phrases representing the intent categories in other domain layers of the intent hierarchy, and wherein the intent hierarchy is independent and different from a hierarchy of the ontology;

determining if the particular intent category has a parent hierarchical intent category located in a higher domain layer of the intent hierarchy; and

if the parent hierarchical intent category is present in the particular intent hierarchy, then displaying an intent response of the parent hierarchical intent category.

2. The computer-readable memory of claim 1, wherein said displaying is according to whether a response tag is set for the particular intent category.

3. The computer-readable memory of claim 1, wherein the single natural language word or phrase for the particular intent category does not exist in any of the queries and is not a natural language query for information.

4. The computer-readable memory of claim 1, wherein the operations further comprise:

analyzing how words in the queries are arranged in sentence structures to identify different natural language meanings; and

automatically generating the single natural language word or phrase for the intent category, the single natural language word or phrase both summarizing and paraphrasing the identified different natural language meanings.

5. The computer-readable memory of claim 1, wherein the operations further comprise:

identifying at least some most frequently queried intent categories for the queries; and

providing pre-query intent responses or links for at least a portion of the identified most frequently queried intent categories.

6. The computer-readable memory of claim 5, wherein the operations further comprise:

identifying a first group of new queries that match the identified most frequently queried intent categories; and

providing pre-configured intent responses for the first group of received new queries corresponding with the matching intent categories.

7. The computer-readable memory of claim 6, wherein the operations further comprise:

identifying a second group of new queries that cannot be matched with the identified most frequently queried intent categories;

using a natural language engine and the same or another ontology to identify concepts in the second group of non-matching new queries;

comparing the identified concepts for the second group of new queries with natural language words or phrases for other intent categories; and

providing predetermined responses to any of the second group of new queries matching any of the other intent categories.

8. The computer-readable memory of claim 7, wherein the operations further comprise:

identifying a third group of new queries that cannot be matched with any of the identified most frequently queried intent categories and cannot be matched with the concepts in the ontologies;

identifying keywords in the third group of new queries; and

using the identified keywords to search for responses associated with the third group of new queries.

9. The computer-readable memory of claim 1, wherein the operations further comprise:

automatically and dynamically creating a single natural language phrase of a plurality of different natural language words describing the particular intent category, wherein automatically and dynamically creating the phrase of the plurality of natural language words comprises;

automatically correcting spelling in the queries;

automatically correcting capitalization and punctuation in the queries;

automatically performing speech analysis in the queries to identify different types of words in the queries;

automatically performing stem analysis on the queries to add other forms of similar words;

automatically performing concept analysis on the queries using the same or another ontology to identify different words that have similar meanings as the words in the queries;

automatically performing linguistic analysis on the queries to identify subject categories for the words in the queries; and

automatically performing intent analysis on the subject categories and concepts identified in the queries to automatically generate the natural language word or phrase summarizing the different natural language formations associated with the queries.

10. An apparatus, comprising:

a processor configured to;

access an ontology that defines a hierarchy of a plurality of concepts at different concept domain layers and that links natural language words to the plurality of concepts at different concept domain layers;

classify each of a plurality of received queries with a particular intent category based on the ontology, wherein the intent category is pre-defined by a single natural language word or phrase that summarizes in a same natural language word formation a generalized information requested by the queries and wherein the queries use different natural language formations to request different types of information summarized by the natural language word or phrase of the particular intent category;

associate an intent response formulated of natural language words or phrases with the particular intent category, wherein the intent response provides a single common applicable response for the queries;

access an intent hierarchy of a plurality of intent categories including the particular intent category, wherein natural language phrases representing the hierarchy of intent categories at different domain layers of the intent hierarchy have different generalities of terms than natural language phrases representing the intent categories in other domain layers of the intent hierarchy, and wherein the intent hierarchy is independent and different from a hierarchy of the ontology;

determine if the particular intent category has a parent hierarchical intent category located in a higher domain layer of the intent hierarchy; and

determine whether to display an additional intent response based on the parent hierarchical intent category determination.

11. The apparatus of claim 10, wherein the processor is configured to operate an Intelligent Matching Language (IML) engine and use the ontology to identify concepts and linguistic characteristics in the queries and compare the concepts and linguistic characteristics of the queries with concepts and linguistic formations of the natural language word formations or phrases representing different intent categories.

12. The apparatus of claim 11, including a memory that stores preconfigured intent responses for only a portion of most frequently queried intent categories.

13. The apparatus of claim 12, wherein the processor is configured to:

receive a new query;

conduct a search using concepts identified using the IML engine and the ontology if no intent category can be identified for the new query; and

conduct a keyword search if no concepts can be identified in the new query.

14. The apparatus of claim 10, wherein the particular intent category includes a parameter identifying the intent response.

15. The apparatus of claim 14, wherein the parameter associates the particular intent category with particular concepts in the ontology.

16. The apparatus of claim 14, wherein the parameter is associated with stored user frequencies information indicating how frequently users use a search engine.

17. The apparatus of claim 10, wherein said display determination is based at least in part on whether a response tag is set for at least one of the intent categories.

18. A method, comprising:

storing an ontology that defines a hierarchy of a plurality of concepts at different concept domain layers and that links natural language words to the plurality of concepts at different concept domain layers;

classifying each of a plurality of received queries with a particular intent category based on the ontology, wherein the intent category is pre-defined by a single natural language word or phrase that summarizes in a same natural language word formation a generalized information requested by the queries and wherein the queries use different natural language formations to request different types of information summarized by the single natural language word or phrase of the particular intent category;

associating, using a computing device, an intent response formulated of natural language words or phrases with the particular intent category, wherein the intent response provides a single common applicable response for the queries;

storing an intent hierarchy of a plurality of intent categories including the particular intent category, wherein natural language phrases representing the hierarchy of intent categories at different domain layers of the intent hierarchy have different generalities of terms than natural language phrases representing the intent categories in other domain layers of the intent hierarchy, and wherein the intent hierarchy is independent and different from a hierarchy of the ontology;

determining, using the computing device, if the particular intent category has a parent hierarchical intent category located in a higher domain layer of the intent hierarchy; and

if the parent hierarchical intent category is present in the intent hierarchy, then using the computer device to cause an intent response of the hierarchical intent category to be displayed.

19. The method of claim 18, wherein said causing is according to whether a response tag is set for the particular intent category.

20. The method of claim 18, further comprising:

analyzing how words in the queries are arranged in sentence structures to identify different natural language meanings; and

automatically generating the single natural language word or phrase for the intent category, the single natural language word or phrase both summarizing and paraphrasing the identified different natural language meanings.