Single unified ranker

Patent metadata

Field	Value
Publication number	US-2017124078-A1
Application number	US-201514925360-A
Country	US
Kind code	A1
Filing date	Oct 28, 2015
Priority date	Oct 28, 2015
Publication date	May 4, 2017
Grant date	—

Abstract

Non-limiting examples of the present disclosure describe a unified ranking model that may be used by a plurality of entry points to return ranked results in response to received query data. The unified ranking model is provided as a service for a plurality of entry points. A query is received from an entry point of the plurality of entry points. Results data for the query data is retrieved. A unified ranking model is executed to rank the results data. Execution of the unified ranking model manipulates feature data of the unified ranking model based on user context signals associated with the received query data and acquired result retrieval signals corresponding with the retrieved results data. Execution of the unified ranking model generates ranked result data. Ranked results data is returned to the processing device corresponding with the entry point. Other examples are also described.

First claim

What is claimed is:

1 . A computer-implemented method comprising:
receiving query data; acquiring, by a processing device, a plurality of user context signals associated with the received query data; retrieving results data for the query data; acquiring, by the processing device, a plurality of result retrieval signals corresponding with the retrieved results data; executing a unified ranking model to rank the results data, wherein the executing comprises:
executing feature normalization on data of the unified ranking model based on the acquired user context signals and the acquired result retrieval signals to generate normalized feature data,
applying the normalized feature data to rank the results data; and
outputting one or more of the ranked results data based on application of the unified ranking model.

2 . The computer-implemented method according to claim 1 , wherein the unified ranking model is provided as a service that is accessible to a plurality of entry points.

3 . The computer-implemented method according to claim 1 , wherein the applying of the normalized feature data further comprises executing machine learning processing that evaluates the normalized feature data to generate the ranked results data.

4 . The computer-implemented method according to claim 3 , wherein the executing of the unified ranking model executes an affinity/distance function to process the acquired user context signals and the acquired result retrieval signals for ranking of the results data.

5 . The computer-implemented method according to claim 1 , wherein the plurality of user context signals comprises signal data corresponding with: location data, language data, location data retrieved from the query data, form-factor data associated with a processing device from which the query data is received, entry point data, time data, and context data associated with a user that submitting the query data.

6 . The computer-implemented method according to claim 1 , wherein the plurality of result retrieval signals comprises signal data corresponding with: location information for the result data, language corresponding with the result data, form-factor of the result data, query click signal data the result data, and timestamp data corresponding to the result data.

7 . The computer-implemented method according to claim 1 , further comprising:
receiving updated query data, acquiring, by the processing device, a plurality of user context signals associated with the updated query data; retrieving updated results data for the updated query data; acquiring, by the processing device, a plurality of result retrieval signals corresponding with the updated results data; executing the unified ranking model to rank the updated results data, and outputting one or more of the updated results data based on application of the unified ranking model.

8 . The computer-implemented method according to claim 1 , wherein the outputting further comprises at least one of: displaying the one or more ranked results on the processing device, and transmitting the one or more ranked results to another processing device.

9 . A system comprising:
at least one processor; and a memory operatively connected with the processor, wherein the memory stores computer-executable instructions, that cause the processor to perform:
providing a unified ranking model as a service for a plurality of entry points;
receiving query data from an entry point of the plurality of entry points, wherein the entry point is associated with a processing device;
retrieving results data for the query data;
executing the unified ranking model to rank the results data, wherein the executing manipulates feature data of the unified ranking model, based on processing of acquired user context signals associated with the received query data and acquired result retrieval signals corresponding with the retrieved results data, and wherein the executing generates ranked results data; and
returning one or more of the ranked results data to the processing device corresponding with the entry point.

10 . The system according to claim 9 , wherein the plurality of user context signals comprises signal data corresponding with: location data, language data, location data retrieved from the query data, form-factor data associated with the processing device associated with the entry point, entry point data, time data, and context data associated with a user that submitted the query data, and wherein the plurality of result retrieval signals comprises signal data corresponding with: location information for the result data, language corresponding with the result data, form-factor for the result data, query click signal data associated with the result data, and timestamp data corresponding to the result data.

11 . The system according to claim 9 , wherein execution of the unified ranking model further comprises executing machine learning processing that evaluates the manipulated feature data based on the acquired user context signals and the acquired result retrieval signals.

12 . The system according to claim 11 , wherein the machine learning processing applies an affinity/distance function to process the acquired user context signals and the acquired result retrieval signals for ranking of the results.

13 . The system according to claim 9 , wherein computer-executable instructions further comprise an instruction that causes the processor to perform:
receiving updated query data from the entry point, acquiring a plurality of user context signals associated with the updated query data; acquiring a plurality of result retrieval signals based on results identified in response to the updated query data; retrieving updated results data for the updated query data; executing the unified ranking model to rank the updated results data; and transmitting one or more of the updated results data to the processing device.

14 . The system according to claim 9 , wherein computer-executable instructions further comprise an instruction that causes the processor to perform:
receiving additional query data from another entry point, wherein the another entry point corresponds with another processing device; acquiring a plurality of user context signals associated with the additional query data; acquiring a plurality of result retrieval signals based on results identified in response to the updated query data; retrieving results data for the additional query data; executing the unified ranking model to rank the results data for the additional query data; and transmitting one or more of the results data to the another processing device corresponding with the another entry point.

15 . A system comprising:
at least one processor; and a memory operatively connected with the processor, wherein the memory stores computer-executable instructions that, when executed, cause the processor to perform:
receiving, from a third-party service, third-party specific data to update a unified ranking model that is configured to rank search results,
providing the updated unified ranking model as a service,
receiving query data from the third-party service,
retrieving results data for the query data;
executing the updated unified ranking model to rank the results data, wherein the executing utilizes the third-party specific data to generate feature data for evaluation of signal data associated with the query data that is received from the third-party service, and
returning at least one of the ranked results data to a processing device corresponding with the third-party service b

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Malik Manish
• Ke Qifa
• Majumder Rangan
• Bode Andreas
• Shukla Pushpraj
• Shi Yu

Classifications

• G06F16/24578 Primary

  using ranking · CPC title

• G06F16/248

  Presentation of query results · CPC title

• G06F16/951

  Indexing; Web crawling techniques · CPC title

• G06F16/24575

  using context · CPC title

• G06F17/3053 Primary

  Physics · mapped topic