Inferring attributes from search queries

• US 8,812,509 B1
• Filed: 11/02/2012
• Issued: 08/19/2014
• Est. Priority Date: 05/18/2007
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Pasca, Alexandru Marius, Van Durme, Benjamin
• Primary Examiner(s): Morrison, Jay
• Assistant Examiner(s): GORTAYO, DANGELINO N
• Application Number:
  US13/667,336
• Time in Patent Office: 655 Days
• Field of Search: 707/1, 707/10, 707/100, 707/104, 707/741, 707/706, 707/732, 707/737, 707/750, 707/780
• US Class Current: 707/741
• CPC Class Codes: G06F 16/33 Querying
• CPC Class Codes: G06F 16/3325 Reformulation based on results of preceding query
• CPC Class Codes: G06F 16/3334 Selection or weighting of terms from queries, including natural language queries
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Systems, techniques, and machine-readable instructions for inferring attributes from search queries. In one aspect, a method includes receiving a description of a collection of search queries, inferring attributes of entities from the description of the collection of search queries, associating the inferred attributes with identifiers of entities characterized by the attributes, and making the associations of the attributes and entities available.

Claims

1. A computer-implemented method comprising:

maintaining a collection of extract patterns, wherein each extract pattern specifies a text pattern that includes an entity placeholder and an attribute placeholder;

maintaining a collection of category definitions, wherein each category definition comprises a category identifier, one or more entity identifiers, and one or more attribute identifiers, each entity identifier comprises an entity-name text string representing a name of an entity in the category, and each attribute identifier comprises an attribute-name text string representing an attribute of the entities in the category;

determining that a first search query text string matches a first extract pattern in the collection of extract patterns, including;

determining that a first substring of the first search query text string matches an entity placeholder in the first extract pattern; and

determining that a second substring of the first search query text string matches an attribute placeholder in the first extract pattern; and

determining that the first substring matches an entity identifier in a first category definition in the collection of category definitions;

adding the second substring as an attribute identifier to a set of one or more first attribute identifiers in the first category definition, the set of one or more first attribute identifiers comprising at least one of the one or more attribute identifiers; and

generating a score for a particular attribute identifier included in the set of one or more first attribute identifiers in the first category definition, wherein the score reflects an extent to which the particular attribute identifier is relevant to characterizing the first category definition.

2. The method of claim 1, further comprising:

removing from the first category definition, after adding the second substring as an attribute identifier to the set of one or more first attribute identifiers in the first category definition, any attribute identifier in the set of one or more first attribute identifiers that identifies a proper name of a person, place, or thing.

3. The method of claim 1, further comprising:

determining that a second search query text string matches a second extract pattern in the collection of extract patterns, including;

determining that a first substring of the second search query text string matches an entity placeholder in the second extract pattern; and

determining that a second substring of the second search query text string matches an attribute placeholder of the second extract pattern; and

determining that the first substring of the second search query matches an entity identifier in the first category definition;

determining that both the second substring of the second search query text string and the second substring of the first search query text string identify a common attribute of an entity within the first category definition and identifying the second substring of the second search query text string and the second substring of the first search query text string as being semantically related to each other; and

determining a weighted frequency with which the common attribute is used to characterize the entity within the first category.

4. The method of claim 1, wherein generating the score comprises:

weighting scores for attributes identified using whole sentence extract patterns more heavily than attributes identified using non-whole sentence extract patterns.

5. The method of claim 1, wherein each extract pattern includes literal text along with the entity placeholder, and the attribute placeholder.

6. The method of claim 5, wherein the entity placeholder and the attribute placeholder are each a respective text string beginning with a “

<

”

character and ending with a “

>

”

character.

7. A system comprising:

one or more computers; and

one or more storage devices in data communication with the one or more computers and storing instructions that, when executed by the one or more computers, cause the one or more computers to perform operations comprising;

maintaining a collection of extract patterns, wherein each extract pattern specifies a text pattern that includes an entity placeholder and an attribute placeholder;

maintaining a collection of category definitions, wherein each category definition comprises a category identifier, one or more entity identifiers, and one or more attribute identifiers, each entity identifier comprises an entity-name text string representing a name of an entity in the category, and each attribute identifier comprises an attribute-name text string representing an attribute of the entities in the category;

determining that a first search query text string matches a first extract pattern in the collection of extract patterns, including;

determining that a first substring of the first search query text string matches an entity placeholder in the first extract pattern; and

determining that a second substring of the first search query text string matches an attribute placeholder in the first extract pattern; and

determining that the first substring matches an entity identifier in a first category definition in the collection of category definitions;

adding the second substring as an attribute identifier to a set of one or more first attribute identifiers in the first category definition, the set of one or more first attribute identifiers comprising at least one of the one or more attribute identifiers; and

generating a score for a particular attribute identifier included in the set of one or more first attribute identifiers in the first category definition, wherein the score reflects an extent to which the particular attribute identifier is relevant to characterizing the first category definition.

8. The system of claim 7, wherein the operations further comprise:

removing from the first category definition, after adding the second substring as an attribute identifier to the set of one or more first attribute identifiers in the first category definition, any attribute identifier in the set of one or more first attribute identifiers that identifies a proper name of a person, place, or thing.

9. The system of claim 7, wherein the operations further comprise:

determining that a second search query text string matches a second extract pattern in the collection of extract patterns, including;

determining that a first substring of the second search query text string matches an entity placeholder in the second extract pattern; and

determining that a second substring of the second search query text string matches an attribute placeholder of the second extract pattern; and

determining that the first substring of the second search query matches an entity identifier in the first category definition;

determining that both the second substring of the second search query text string and the second substring of the first search query text string identify a common attribute of an entity within the first category definition and identifying the second substring of the second search query text string and the second substring of the first search query text string as being semantically related to each other; and

determining a weighted frequency with which the common attribute is used to characterize the entity within the first category.

10. The system of claim 7, wherein the operations of generating the score comprise:

weighting scores for attributes identified using whole sentence extract patterns more heavily than attributes identified using non-whole sentence extract patterns.

11. The system of claim 7, wherein each extract pattern includes literal text along with the entity placeholder, and the attribute placeholder.

12. The system of claim 11, wherein the entity placeholder and the attribute placeholder are each a respective text string beginning with a “

<

”

character and ending with a “

>

”

character.

13. A computer program product, encoded on one or more non-transitory computer storage media, comprising instructions that, when executed by one or more computers, cause the one or more computers to perform operations comprising:

maintaining a collection of extract patterns, wherein each extract pattern specifies a text pattern that includes an entity placeholder and an attribute placeholder;

maintaining a collection of category definitions, wherein each category definition comprises a category identifier, one or more entity identifiers, and one or more attribute identifiers, each entity identifier comprises an entity-name text string representing a name of an entity in the category, and each attribute identifier comprises an attribute-name text string representing an attribute of the entities in the category;

determining that a first search query text string matches a first extract pattern in the collection of extract patterns, including;

determining that a first substring of the first search query text string matches an entity placeholder in the first extract pattern; and

determining that a second substring of the first search query text string matches an attribute placeholder in the first extract pattern; and

determining that the first substring matches an entity identifier in a first category definition in the collection of category definitions;

adding the second substring as an attribute identifier to a set of one or more first attribute identifiers in the first category definition, the set of one or more first attribute identifiers comprising at least one of the one or more attribute identifiers; and

generating a score for a particular attribute identifier included in the set of one or more first attribute identifiers in the first category definition, wherein the score reflects an extent to which the particular attribute identifier is relevant to characterizing the first category definition.

14. The product of claim 13, wherein the operations further comprise:

removing from the first category definition, after adding the second substring as an attribute identifier to the set of one or more first attribute identifiers in the first category definition, any attribute identifier in the set of one or more first attribute identifiers that identifies a proper name of a person, place, or thing.

15. The product of claim 13, wherein the operations further comprise:

determining that a second search query text string matches a second extract pattern in the collection of extract patterns, including;

determining that a first substring of the second search query text string matches an entity placeholder in the second extract pattern; and

determining that a second substring of the second search query text string matches an attribute placeholder of the second extract pattern; and

determining that the first substring of the second search query matches an entity identifier in the first category definition;

determining that both the second substring of the second search query text string and the second substring of the first search query text string identify a common attribute of an entity within the first category definition and identifying the second substring of the second search query text string and the second substring of the first search query text string as being semantically related to each other; and

determining a weighted frequency with which the common attribute is used to characterize the entity within the first category.

16. The product of claim 13, wherein the operations for generating the score comprise:

weighting scores for attributes identified using whole sentence extract patterns more heavily than attributes identified using non-whole sentence extract patterns.

17. The product of claim 13, wherein each extract pattern includes literal text along with the entity placeholder, and the attribute placeholder.

18. The product of claim 17, wherein the entity placeholder and the attribute placeholder are each a respective text string beginning with a “

<

”

character and ending with a “

>

”

character.