Query pattern generation for answers coverage expansion

• US 8,515,986 B2
• Filed: 12/02/2010
• Issued: 08/20/2013
• Est. Priority Date: 12/02/2010
• Status: Active Grant

• Current Assignee: Zhigu Holdings Limited
• Original Assignee: Microsoft Corporation
• Inventors: Salvetti, Franco, Tu, Ying, Ahn, David D
• Primary Examiner(s): BURKE, JEFF A
• Application Number:
  US12/959,050
• Publication Number:
  US 20120143895A1
• Time in Patent Office: 992 Days
• Field of Search: 707/765, 707/768
• US Class Current: 707/768
• CPC Class Codes: G06F 16/3329 Natural language query formulation or dialogue systems

Abstract

Answers are provided to users in response to queries as a supplement to any responsive documents. Query formats for entity and attribute combinations are identified. The query formats can be substituted with entity and attribute combinations that have a corresponding attribute value to form a list of answered queries. The attribute value corresponding to an answered query can be provided when a query is received that matches an answered query.

Claims

1. A method for associating user queries with answers, comprising:

providing a processor and a memory containing computer executable instructions; and

executing, by the processor, the computer executable instructions to provide a method comprising;

identifying an attribute having an attribute value for a plurality of corresponding entities;

constructing basic queries based on combinations of the attribute with the plurality of corresponding entities;

determining frequencies for the constructed basic queries in a first query logfile;

selecting one or more entities based on the determined frequencies;

extracting queries from a second query logfile, the extracted queries containing the identified attribute and a selected entity;

generating one or more query formats by generalizing the selected entities in the extracted queries;

receiving an input to remove a plurality of the generated query formats; and

forming a list of answered queries by substituting the generalized entities in the one or more query formats with the plurality of entities corresponding to the identified attribute, each answered query corresponding to an attribute value associated with the entity and the attribute in the query.

2. The method of claim 1, further comprising:

receiving a query;

matching the received query with an answered query; and

displaying the attribute value corresponding to the answered query.

3. The method of claim 1, further comprising:

receiving a query;

matching the received query with an answered query;

forwarding at least the attribute and the entity from the received query to an answer database;

receiving an attribute value; and

displaying the attribute value corresponding to the answered query.

4. The method of claim 1, wherein the first query logfile is the same as the second query logfile.

5. The method of claim 1, wherein an extracted query containing the identified attribute comprises an extracted query containing an attribute expression corresponding to the identified attribute.

6. The method of claim 1, wherein generating one or more query formats comprises replacing the selected entities with a placeholder expression.

7. The method of claim 1, wherein the plurality of corresponding entities includes variants for at least one of the entities.

8. The method of claim 1, further comprising filtering the generated query formats based on frequencies determined from the second query logfile.

9. The method of claim 1, wherein the generated query formats are filtered based on at least one of a total number of queries corresponding to the generated query format and a number of distinct queries corresponding to the generated query format.

10. The method of claim 1, wherein determining frequencies of basic queries in a first query logfile comprises:

determining a frequency for each basic query, wherein at least two basic queries contain variants of an entity; and

aggregating the frequency for the basic queries containing the variants of the entity.

11. The method of claim 1, wherein the input to remove the plurality of generated query formats comprises input from a person reviewing the generated query formats.

12. A method for associating user queries with answers, comprising:

providing a processor and a memory containing computer executable instructions; and

executing, by the processor, the computer executable instructions to provide a method comprising;

identifying an attribute having an attribute value for one or more corresponding entities;

extracting queries from a query logfile, the extracted queries containing the identified attribute and at least one corresponding entity;

generating one or more query formats by generalizing the entities in the extracted queries;

receiving an input to remove a plurality of the generated query formats;

forming a list of answered queries by substituting the generalized entities in the one or more query formats with the one or more entities corresponding to the identified attribute, each answered query corresponding to an answer associated with the entity and the attribute in the query;

receiving a query corresponding to an answered query; and

displaying the attribute value corresponding to the answered query.

13. The method of claim 12, wherein an extracted query containing the identified attribute comprises an extracted query containing an attribute expression corresponding to the identified attribute.

14. The method of claim 12, further comprising matching the received query with an answered query.

15. The method of claim 12, further comprising filtering the generated query formats based on frequencies determined from the query logfile.

16. The method of claim 15, wherein the generated query formats are filtered based on at least one of a total number of queries corresponding to the generated query format and a number of distinct queries corresponding to the generated query format.

17. The method of claim 12, wherein generating one or more query formats comprises replacing the entities with a placeholder expression.

18. A method for associating user queries with answers, comprising:

identifying an attribute having an attribute value for one or more corresponding entities, the attribute being represented by one or more attribute expressions;

extracting queries from a query logfile, the extracted queries containing an attribute expression corresponding to the identified attribute and at least one corresponding entity;

generating one or more query formats by generalizing the entities and the attribute expressions in the extracted queries;

receiving an input to remove a plurality of the generated query formats;

filtering the generated query formats based on frequencies determined from the query logfile; and

forming a list of answered queries by substituting the generalized attribute with an attribute expression and substituting the generalized entity with an entity corresponding to the attribute for the attribute expression, each answered query corresponding to an attribute value associated with the entity and the attribute in the query.

19. The method of claim 18, wherein the at least one corresponding entity in the extracted queries is selected based on a frequency of the at least one corresponding entity in the query logfile.

20. The method of claim 19, further comprising:

constructing basic queries based on combinations of the identified attribute with the one or more corresponding entities;

determining frequencies for the basic queries in the query logfile; and

selecting the at least one corresponding entity based on the determined frequencies.