Systems and methods for data mining and automated generation of search query rewrites

• US 10,599,733 B2
• Filed: 03/30/2017
• Issued: 03/24/2020
• Est. Priority Date: 12/22/2014
• Status: Active Grant

• Current Assignee: eBay Inc.
• Original Assignee: eBay Inc.
• Inventors: Kumar, Prathyusha Senthil, Arasada, Praveen, Jammalamadaka, Ravi Chandra
• Primary Examiner(s): Nguyen, Cam Linh T
• Application Number:
  US15/474,560
• Publication Number:
  US 20170206280A1
• Time in Patent Office: 1,090 Days
• Field of Search: 707706, 707708, 707721
• CPC Class Codes: G06F 16/24534 Query rewriting; Transformation
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/953 Querying, e.g. by the use of web search engines
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9538 Presentation of query results
• CPC Class Codes: G06F 16/955 using information identifiers, e.g. uniform resource locators [URL]

Abstract

Various embodiments include systems and methods for data mining of search engine and network operations to automatically identify query events. Data aggregated from such query events and stored as query history data may be processed to identify query ranking mismatches. These identified mismatches may be used with the query history data and target settings to automatically generate query rewrite data. In certain embodiments, this query rewrite data may be used to generate query rewrite input language (QRIL) records. Such QRIL records may then be used to automatically generate standardized rewrites which automatically resolve any conflicts between rewrites in a particular search engine.

Claims

1. A query rewrite system comprising:

a data mining device comprising;

an input module configured to receive, from a network system comprising a search engine, a set of query history data, the set of query history data comprising at least a query value, a set of ranked search results associated with the query value, and a set of user interaction data associated with the query value; and

a data mining module coupled to the input module configured to automatically process the set of query history data to identify a query ranking mismatch and to generate a set of query rewrite data, wherein the set of query rewrite data comprises constraint data and rewrite data, wherein the constraint data comprises at least a first trigger value, and wherein the rewrite data identifies at least a first query rewrite value associated with the first trigger value.

2. The system of claim 1 wherein the data mining module comprises a data mining system setting module comprising a first target setting, the first target setting identifying a first query type and a set of target ranked result characteristics associated with the first query type.

3. The system of claim 2 wherein the data mining module further comprises a ranking mismatch identification module, the ranking mismatch identification module configured to:

identify query history events in a query history database associated with the first query type;

determine a first set of ranked result characteristics associated with the query history events;

determine that the first set of ranked result characteristics do not meet the target ranked result characteristics; and

identify the query ranking mismatch based on the determination that the first set of ranked result characteristics to not meet the target ranked result characteristics.

4. The system of claim 3 wherein the first target setting further comprises the first trigger value and the first query rewrite value.

5. The system of claim 4 wherein the data mining system setting module is configured to generate the set of query rewrite data in response to the ranking mismatch identification module determining that the first set of ranked result characteristics do not meet the target ranked result characteristics.

6. The system of claim 5 wherein the data mining module comprises a ranking model associated with a ranking process of the search engine, wherein the data mining system setting module is further configured to generate the query rewrite data in response to a determination by the ranking model that one or more standardized rewrites associated with the query rewrite data creates ranked search results associated with the first query type that meet the target ranked result characteristics.

7. The system of claim 5 wherein the data mining module further comprises a merchant customization module that provides the first target setting to the data mining system setting module;

and

wherein the query rewrite data comprises a constraint value restricting the query rewrite data to a first category of a category tree associated with the search engine.

8. The system of claim 5 wherein the data mining module comprises a category customization module that provides the first target setting to the data mining system setting module;

and

wherein the query rewrite data comprises a constraint value restricting the query rewrite data to a storefront of a marketplace system associated with a first merchant.

9. The system of claim 3 wherein the data mining module comprises a query rewrite history module configured to:

parse the set of query history data to identify a query history event associated with the query value, the set of ranked search results, and the set of user interaction data associated with the query value;

associate the query type with the query history event; and

store the query value, the set of ranked search results, the set of user interaction data, and the query type as a first query history event of the query history events in the query history event database.

10. The system of claim 1 further comprising:

a query transcoding device configured to receive, from the automatic query rewrite generation device, the set of query rewrite data;

wherein the query transcoding device generates a first query rewrite input language (QRIL) record from the set of query rewrite data, wherein the first QRIL record comprises the first trigger value, the first query rewrite value, and a first metaflag element that identifies the first QRIL record as associated with the first query rewrite type.

11. The system of claim 10 further comprising:

a QRIL record database coupled to the query transcoding device;

a QRIL processor device coupled to the QRIL record database, the QRIL processor device configured to access the QRIL record database to retrieve the first QM!, record, process the first QRIL record to generate a first standardized rewrite, and updating a set of standardized rewrites using the first standardized rewrite to generate an updated set of standardized rewrites.

12. The system of claim 10 further comprising:

the search engine system coupled to the QRIL processor, wherein the search engine system receives the updated set of standardized rewrites, rewrites a first search engine user query from a first client device using the first standardized rewrite to create a rewritten user query, generates a first set of search results for the using the rewritten user query, and communicates the first set of search results the search engine to the first client device.

13. The system of claim 12 wherein the search engine communicates the updated set of standardized rewrites, the first search engine user query, and the first set of search results to the data mining device as second set of query history data.

14. A method of automatically generating query rewrites comprising:

receiving, at an input module of a data mining device, a set of query history data;

processing, using at least a first processor, the set of query history to identify a query ranking mismatch associated with the set of query history data; and

generating, in response to the identification of the query ranking mismatch, a first set of query rewrite data comprising a first trigger value and a first query rewrite value associated with the first trigger value.

15. The method of claim 14 further comprising:

parsing the set of query history data to identify a query history event associated with the query value, the set of ranked search results, and the set of user interaction data associated with the query value;

associating a query history type with the query history event; and

storing the query value, the set of ranked search results, the set of user interaction data, and the query history type as a first query history event in the query history event database.

16. The method of claim 15 wherein processing the set of query history to identify the query ranking mismatch comprises:

receiving a first target setting, the first target setting identifying the first query history type and a set of target ranked result characteristics associated with the first query history type;

identifying a plurality of query history events comprising the first query history event in a query history database, the plurality of query history events being associated with the first query history type;

determining a first set of ranked result characteristics associated with the plurality of query history events;

determining that the first set of ranked result characteristics do not meet the target ranked result characteristics; and

identifying the query ranking mismatch based on the determination that the first set of ranked result characteristics to not meet the target ranked result characteristics.

17. The method of claim 16 wherein generating the first set of query rewrite data comprises:

identifying a difference between the first set of ranked result characteristics and the target ranked result characteristics;

determining that the first trigger value and the first query rewrite value are associated with the difference between the first set of ranked result characteristics and the target ranked result characteristics; and

generating the first set of query rewrite data using the first query rewrite value and the first trigger value based on the determination that the first trigger value and the first query rewrite value are associated with the difference between the first set of ranked result characteristics and the target ranked result characteristics.

18. The method of claim 16 further comprising:

analyzing the first set of query data to identify a first query rewrite type associated with the first set of query rewrite data from a plurality of query rewrite types;

generating a first query rewrite input language (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 a first metaflag element that identifies the first QRIL record as associated with the first query rewrite type;

storing the first QRIL record in a QRIL record database with a plurality of QRIL records;

accessing, by a QRIL processor module, the QRIL record database to retrieve the first QRIL record;

processing the first QRIL record using the QRIL processor module to generate a first standardized rewrite; and

updating a set of standardized rewrites using the first standardized rewrite to generate an updated set of standardized rewrites;

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

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;

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

communicating the first search engine user query and the first set of search results to the data mining device as a second set of query history data.

19. A non-transitory computer readable medium comprising computer readable instructions that, when executed by a device comprising at least one processor and a memory coupled to the at least one processor, cause the device to:

receive a set of query history data;

process the set of query history to identify a query ranking mismatch associated with the set of query history data; and

generate, in response to the identification of the query ranking mismatch, a first set of query rewrite data comprising a first trigger value and a first query rewrite value associated with the first trigger value.