公开(公告)号：US10459979B2

公开(公告)日：2019-10-29

申请号：US15199335

申请日：2016-06-30

Applicant: Facebook, Inc.

Inventor： Szymon Piechowicz ,  Barak Reuven Naveh ,  Annie Hsin-Wen Liu ,  Ashish Gupta

IPC: G06F15/16 ,  G06F16/901 ,  G06F16/28 ,  G06Q10/04 ,  G06Q50/00

Abstract: Various embodiments include a classification platform system. A user can define a classification experiment on the classification platform system. For example, the user can define an input data space by selecting at least one of data sources interfaced with the classification platform system and defining a workflow configuration including a directed graph (DG) connecting a plurality of transformation blocks to represent an experiment workflow. The DG can specify how one or more outputs of each of the transformation blocks are fed into one or more other transformation blocks. The DG can be defined graphically. The classification platform system can schedule the experiment workflow to be executed on a distributed computation platform according to the input data space and the workflow configuration.

公开(公告)号：US20180004859A1

公开(公告)日：2018-01-04

申请号：US15199351

申请日：2016-06-30

Applicant: Facebook, Inc.

Inventor： Szymon Piechowicz ,  Barak Reuven Naveh ,  Annie Hsin-Wen Liu ,  Ashish Gupta

IPC: G06F17/30

Abstract: Various embodiments include a classification platform system. A user can define a classification experiment on the classification platform system. For example, the user can define an input data space by selecting at least one of data sources interfaced with the classification platform system and defining a workflow configuration including a directed graph (DG) connecting a plurality of transformation blocks to represent an experiment workflow. The DG can specify how one or more outputs of each of the transformation blocks are fed into one or more other transformation blocks. The DG can be executed by various types of computation platforms. The classification platform system can schedule the experiment workflow to be executed on a distributed computation platform according to the input data space and the workflow configuration.

公开(公告)号：US20180004835A1

公开(公告)日：2018-01-04

申请号：US15199403

申请日：2016-06-30

Applicant: Facebook, Inc.

Inventor： Szymon Piechowicz ,  Barak Reuven Naveh ,  Annie Hsin-Wen Liu ,  Ashish Gupta

IPC: G06F17/30 ,  G06N99/00

CPC classification number: G06N20/00 ,  G06F16/20 ,  G06N5/022

Abstract: Various embodiments include a classification platform system. A user can define a classification experiment on the classification platform system. For example, the user can define an input data space by selecting at least one of data sources interfaced with the classification platform system and defining a directed graph (DG) connecting a plurality of transformation blocks to represent an experiment workflow. The DG can specify how one or more outputs of each of the transformation blocks are fed into one or more other transformation blocks. At least one of the transformation block can dynamically modify the DG. The classification platform system can schedule the experiment workflow to be executed on a distributed computation platform according to the input data space and the DG.

公开(公告)号：US20200334293A1

公开(公告)日：2020-10-22

申请号：US16916040

申请日：2020-06-29

Applicant: FACEBOOK, INC.

Inventor： Szymon Piechowicz ,  Barak Reuven Naveh ,  Annie Hsin-Wen Liu ,  Ashish Gupta

IPC: G06F16/901

Abstract: Various embodiments include a classification platform system. A user can define a classification experiment on the classification platform system. For example, the user can define an input data space by selecting at least one of data sources interfaced with the classification platform system and defining a workflow configuration including a directed graph (DG) connecting a plurality of transformation blocks to represent an experiment workflow. The DG can specify how one or more outputs of each of the transformation blocks are fed into one or more other transformation blocks. The DG can be executed by various types of computation platforms. The classification platform system can schedule the experiment workflow to be executed on a distributed computation platform according to the input data space and the workflow configuration.

公开(公告)号：US20180007145A1

公开(公告)日：2018-01-04

申请号：US15199335

申请日：2016-06-30

Applicant: Facebook, Inc

Inventor： Szymon Piechowicz ,  Barak Reuven Naveh ,  Annie Hsin-Wen Liu ,  Ashish Gupta

IPC: H04L29/08 ,  H04L12/26

CPC classification number: G06F16/9024 ,  G06F16/285 ,  G06Q10/04 ,  G06Q50/01

Abstract: Various embodiments include a classification platform system. A user can define a classification experiment on the classification platform system. For example, the user can define an input data space by selecting at least one of data sources interfaced with the classification platform system and defining a workflow configuration including a directed graph (DG) connecting a plurality of transformation blocks to represent an experiment workflow. The DG can specify how one or more outputs of each of the transformation blocks are fed into one or more other transformation blocks. The DG can be defined graphically. The classification platform system can schedule the experiment workflow to be executed on a distributed computation platform according to the input data space and the workflow configuration.

公开(公告)号：US10698954B2

公开(公告)日：2020-06-30

申请号：US15199351

申请日：2016-06-30

Applicant: Facebook, Inc.

Inventor： Szymon Piechowicz ,  Barak Reuven Naveh ,  Annie Hsin-Wen Liu ,  Ashish Gupta

IPC: G06F16/901

Abstract: Various embodiments include a classification platform system. A user can define a classification experiment on the classification platform system. For example, the user can define an input data space by selecting at least one of data sources interfaced with the classification platform system and defining a workflow configuration including a directed graph (DG) connecting a plurality of transformation blocks to represent an experiment workflow. The DG can specify how one or more outputs of each of the transformation blocks are fed into one or more other transformation blocks. The DG can be executed by various types of computation platforms. The classification platform system can schedule the experiment workflow to be executed on a distributed computation platform according to the input data space and the workflow configuration.

公开(公告)号：US10102255B2

公开(公告)日：2018-10-16

申请号：US15260214

申请日：2016-09-08

Applicant: Facebook, Inc.

Inventor： Dinkar Jain ,  Ashish Gupta ,  Rose Marie Philip ,  Rajat Raina

IPC: G06F17/30

Abstract: In one embodiment, a method includes receiving a query inputted by the user; identifying a set of objects matching the query; calculating, for each identified object, a plurality of category-scores corresponding to a plurality of categories, respectively, wherein each category-score is calculated based on a plurality of sub-scores corresponding to a plurality of scoring axes; categorizing each identified object into a category of the plurality of categories based on the category-scores for the identified object; and sending, to the client system in response to the query, one or more search results corresponding to one or more of the categorized objects for display, each search result referencing the respective categorized object, wherein the one or more categorized objects of the search results comprises objects categorized into one or more selected categories.

公开(公告)号：US10776437B2

公开(公告)日：2020-09-15

申请号：US15702558

申请日：2017-09-12

Applicant: Facebook, Inc.

Inventor： Ashish Gupta ,  Jiajin Yu ,  Rajat Raina

IPC: G06F16/9535 ,  G06F16/248 ,  G06F16/2458 ,  G06F16/2457 ,  G06F16/2453 ,  G06F16/242

Abstract: In one embodiment, a method includes receiving a search query and identifying one or more n-grams. The method also includes accessing, for each identified n-gram, a time-based distribution that includes a count of actions associated with the n-gram. The actions may have occurred within several time windows, and the time-based distribution may count the actions within each time window. The method may also include determining one or more extrema time windows. The method may also include modifying the search query based on the determined extrema time windows. The method also includes executing the modified search query, ranking identified content objects, and sending search results to a client system. The search results may be displayed in ranked order based on the rankings of the corresponding content objects.

公开(公告)号：US20190354554A1

公开(公告)日：2019-11-21

申请号：US16530587

申请日：2019-08-02

Applicant: Facebook, Inc.

Inventor： Szymon Piechowicz ,  Barak Reuven Naveh ,  Annie Hsin-Wen Liu ,  Ashish Gupta

IPC: G06F16/901 ,  G06Q50/00 ,  G06F16/28 ,  G06Q10/04

Abstract: In one embodiment, a method includes defining an input data space, defining a directed graph (DG) connecting a plurality of transformation blocks to represent an experiment workflow, wherein at least one of the plurality of transformation blocks includes logic to dynamically modify the DG during execution of the experiment workflow, formatting the DG and the input data space into a data structure such that the data structure is interpretable by a plurality of different computation platforms, scheduling a distributed computation platform selected from the plurality of different computation platforms to execute the experiment workflow according to the input data space and the DG, and imperatively programming computing nodes of the distributed computation platform to execute the experiment workflow based on the DG if at least one of the transformation blocks dynamically modifies the DG during the execution of the experiment workflow.

公开(公告)号：US20190080010A1

公开(公告)日：2019-03-14

申请号：US15702558

申请日：2017-09-12

Applicant: Facebook, Inc.

Inventor： Ashish Gupta ,  Jiajin Yu ,  Rajat Raina

IPC: G06F17/30

Abstract: In one embodiment, a method includes receiving a search query and identifying one or more n-grams. The method also includes accessing, for each identified n-gram, a time-based distribution that includes a count of actions associated with the n-gram. The actions may have occurred within several time windows, and the time-based distribution may count the actions within each time window. The method may also include determining one or more extrema time windows. The method may also include modifying the search query based on the determined extrema time windows. The method also includes executing the modified search query, ranking identified content objects, and sending search results to a client system. The search results may be displayed in ranked order based on the rankings of the corresponding content objects.