Systems and methods for generating search query rewrites

• US 10,108,712 B2
• Filed: 11/19/2014
• Issued: 10/23/2018
• Est. Priority Date: 11/19/2014
• Status: Active Grant

• Current Assignee: eBay Inc.
• Original Assignee: eBay Inc.
• Inventors: Kumar, Prathyusha Senthil, Arasada, Praveen, Jammalamadaka, Ravi Chandra
• Primary Examiner(s): Morrison, Jay A
• Application Number:
  US14/548,126
• Publication Number:
  US 20160140233A1
• Time in Patent Office: 1,434 Days
• CPC Class Codes: G06F 16/24534 Query rewriting; Transformation
• CPC Class Codes: G06F 16/24565 Triggers; Constraints
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Various embodiments include systems and methods tier processing query rewrite records to generate standardized query rewrites usable by a search engine. Such systems and method may involve analysis of query rewrite input language (QRIL) records to identify relationships and conflicts between multiple QRIL records, and to resolve these relationships and conflicts to generate a standardized rewrite in a semantic language recognizable by the search engine. Such systems and methods may gather QRIL records from a QRIL record database, process the QRIL records using precedence rules, and then communicate a set of standardized and optimized query rewrites to the search engine.

Claims

1. A system comprising:

a query rewrite input language (QRIL) processor device comprising;

an input/output (I/O) module configured to receive one or more QRIL records, each QRIL record comprising a trigger value, a rewrite value, and one or more metaflag elements comprising a query rewrite type metaflag that identifies each QRIL record as associated with a query rewrite type;

a query rewrite resolver module configured to identify a first query rewrite type for a first QRIL record of the one or more QRIL records and process the first QRIL record to identify a set of precedence issues between the first QRIL record and at least one overlapping or conflicting QRIL record; and

a search engine query rewrite customization module configured to identify a first search engine and a first query rewrite semantic structure associated with the first search engine and to generate a first standardized query rewrite from the first QRIL record, the set of precedence issues, and the first query rewrite semantic structure, wherein the first standardized query rewrite comprises a standardized trigger value and a standardized rewrite expression for rewriting an input query for the first search engine, and a bag of words dictionary;

wherein the QRIL processor device is configured to provide the first standardized query write to facilitate executing a user query using the first standardized query write.

2. The system of claim 1 wherein the QRIL processor device further comprises a segmenter module configured to analyze tokens of a first trigger value and a first rewrite value of the first QRIL record to generate value segments and to identify a second QRIL record of the one or more QRIL records associated with one of the value segments.

3. The system of claim 2 wherein the QRIL processor device further comprises a fuzzy query rewrite generator module configured to process the first QRIL record and the second QRIL record that is different than the first QRIL record to generate a standardized rewrite comprising the standardized trigger value derived from the first trigger value and the standardized rewrite expression derived from the first rewrite value and a second rewrite value of the second QRIL record.

4. The system of claim 1 wherein the QRIL processor device further comprises a bag of words generator module comprising a bag of words model configured to analyze tokens of the first QRIL record to identify dictionary values associated with the first QRIL record;

and

wherein the first standardized query rewrite is generated using the dictionary values associated with the first QRIL record.

5. The system of claim 1 wherein the QRIL processor device further comprises a query rewrite optimizer module configured to receive the first standardized query rewrite and apply Boolean logic to generate an optimized first standardized query rewrite that consists of fewer characters than the first standardized query rewrite.

6. The system of claim 1 further comprising:

a QRIL record database configured to store the one or more QRIL, records and communicate the one or more QRIL records to the QRIL processor device; and

a search engine system coupled to the QRIL processor device, wherein the search engine system is configured to receive the first standardized query rewrite from the QRIL processor, query rewrite a first search engine user query from a first client device using the first standardized query rewrite to create a rewritten user query, generate a first set of search results for the using the rewritten user query, and communicate the first set of search results the search engine to the first client device.

7. The system of claim 6 further comprising:

a storage memory configured to store a set of standardized rewrites comprising the first standardized query rewrite, and configured to communicate the set of standardized rewrites to the search engine after the QRIL processor device has processed each QRIL record of the one or more QRIL records.

8. The system of claim 7 further comprising:

a query transcoding device coupled to the QRIL record database, wherein the query transcoding device comprises;

an input module configured to receive, from a first query rewrite source device, a first set of query rewrite data, wherein the first set of query rewrite data comprises constraint data, metaflag data, and rewrite data, wherein the constraint data comprises at least the first trigger value, and wherein the rewrite data identifies at least the first query rewrite value associated with the first trigger value;

a data parser module coupled to the input module configured to process the first set of query data to identify the first trigger and the first query rewrite value, and to communicate parsed query data to one or more identifier module to identify the first query rewrite type associated with the first set of query rewrite data from a plurality of query rewrite types;

a query rewrite input language (QRIL) record generation and formatting module configured to generate the first QRIL record from the first set of query rewrite data, wherein the first QRIL record comprises the first trigger value, the first query rewrite value, and the query rewrite type metaflag, and wherein the QRIL record generation and formatting module is configured to generate a category constraint value, a merchant site constraint value, and a query origination country value as part of the first QRIL record based on the analyzing of the query rewrite data by the constraint identifier module; and

a constraint identifier module coupled to the data parser module configured to analyze the first set of query rewrite data to identify one or more constraints associated with application of the first trigger value to a user query.

9. A method comprising:

receiving at an input module of a query rewrite input language (QRIL) processor device, a one or more QRIL records, each QRIL record comprising a trigger value, a query rewrite value, and one or more metaflag elements comprising a query rewrite type metaflag that identifies each QRIL record as associated with a query rewrite type, wherein the QRIL processor device comprises a memory and one or more processors coupled to the memory;

identifying, by a rewrite resolver module, a first query rewrite type for a first QRIL record of the one or more QRIL records;

processing, by the rewrite resolver module, using the first query rewrite type, the first QRIL record to identify a set of precedence issues between the first QRIL record and at least one overlapping or conflicting QRIL record;

identifying, by a search engine rewrite customization module, a first search engine and a first rewrite semantic structure associated with the first search engine;

generating, by the search engine rewrite customization module, a first standardized rewrite from the first QRIL record, the set of precedence issues, and the first rewrite semantic structure, wherein the first standardized rewrite comprises a standardized trigger value, a standardized rewrite expression for rewriting an input query for the first search engine, and a bag of words dictionary; and

providing, by the QRIL processor device, the first standardized rewrite to facilitate executing a user query using the first standardized rewrite.

10. The method of claim 9 further comprising:

analyzing, using a segmenter module, a first trigger value of the first QRIL to identify one or more trigger segments; and

analyzing, using the segmenter module, the one or more QRIL records to identify trigger values that match at least one of the one or more trigger segments.

11. The method of claim 10 further comprising:

identifying, using the segmenter module, a first trigger segment of the first trigger value, wherein the first trigger value comprises a plurality of tokens and wherein the first trigger segment consists of a first token of the plurality of tokens;

identifying, using the segmenter module, a second QRIL record, wherein a second trigger value of the second QRIL, record consists of the first token;

wherein the first standardized rewrite is further generated using the second QRIL record.

12. The method of claim 11 further comprising:

generating, using a fuzzy rewrite generator module a fuzzy rewrite value associated with the first QRIL, where the fuzzy rewrite value is generated from the first QRIL record and the second QRIL record; and

wherein the first standardized rewrite comprises a rewrite expression that is the fuzzy rewrite value expressed in the first rewrite semantic structure.

13. The method of claim 9 further comprising:

parsing the first QRIL record with a bag of words generator module using a category tree dictionary to generate a first dictionary token list for the first QRII record;

wherein the first standardized rewrite comprises the first dictionary token list.

14. The method of claim 9 further comprising:

analyzing, using a rewrite optimizer module, the first standardized rewrite to identify a first logical structure of the first standardized rewrite as expressed in the first rewrite semantic structure;

identifying a second logical structure different from the first logical structure and that is logically equivalent to the first logical structure; and

generating an optimized first standardized rewrite comprising the second logical structure expressed in the first rewrite semantic structure using the first standardized rewrite.

15. The method of claim 14 wherein the first standardized rewrite comprises at least two search commands associated with a first category of a category tree;

and

wherein the optimized first standardized rewrite consists of a single search command associated with the first category of the category tree.

16. The method of claim 9 further comprising:

receiving, at a query factorization module of a search engine, an updated set of standardized rewrites comprising the first standardized rewrite;

receiving, at the search engine from a first client device, a first search engine user query;

rewriting the first search engine user query using the first standardized rewrite and the query factorization module;

generating a first set of search results for the first search engine user query using the first standardized rewrite; and

communicating the first set of search results for the first search engine user query from the search engine to the first client device.

17. A non-transitory computer readable medium comprising computer readable instructions that, when executed by a device, cause the device to:

receive one or more QRIL records, each QRIL record comprising a trigger value, a query rewrite value, and one or more metaflag elements comprising a query rewrite type metaflag that identifies each QRIL record as associated with a query rewrite type;

identify a first query rewrite type for a first QRIL record of the one or more QRIL, records;

process, using the first query rewrite type, the first QRIL record to identify a set of precedence issues between the first QRIL record and at least one overlapping or conflicting QRIL record;

identify a first search engine and a first rewrite semantic structure associated with the first search engine; and

generate a first standardized rewrite from the first QRIL record, the set of precedence issues, and the first rewrite semantic structure, wherein the first standardized rewrite comprises a standardized trigger value and a standardized rewrite expression for rewriting an input query for the first search engine, and a bag of words dictionary; and

provide the first standardized rewrite to facilitate executing a user query using the first standardized rewrite.

18. The non-transitory computer readable instruction medium of claim 17 wherein processing the first QRIL record to identify a set of precedence issues associated with the first QRIL record comprises:

identifying a second QRIL record, wherein a first set of tokens of a first trigger of the first QRIL record are the same as a second set of tokens of a second trigger of the second QRIL record and wherein a first query rewrite type associated with the first QRIL record is different than a second query rewrite type associated with the second QRIL record.

19. The non-transitory computer readable instruction medium of claim 18 wherein the first Query rewrite type consists of one of:

a phrase rewrite type;

a token refinement type;

a direct rewrite type; and

a whole query rewrite type.

20. The non-transitory computer readable instruction medium of claim 19 wherein generating the first standardized rewrite from the first QRIL record, the set of precedence issues, and the first rewrite semantic structure comprises:

generating the first standardized rewrite using a second rewrite value of the second QRIL record without using a first rewrite value of the first QRIL record based on a type priority of the second query type over the first query type.