公开(公告)号：US10671927B1

公开(公告)日：2020-06-02

申请号：US15662046

申请日：2017-07-27

Applicant: Google Inc.

Inventor： Yifang Liu

IPC: G06N5/02

Abstract: The modeling of an impression effect may include generating a content item impression effect distribution. A classification model may be used to determine a period of the content item impression effect distribution based on one or more accessed impression effect parameters. A value for a content item may be determined based, at least in part, on the determined period and a bid associated with the content item. A content item may be selected based on the determined value and data to display the selected content item may be transmitted. In some instances, the determined period may be used to determine or select predictive model for the determined period that outputs a factor to modify the determined value.

公开(公告)号：US09817880B1

公开(公告)日：2017-11-14

申请号：US14095723

申请日：2013-12-03

Applicant: GOOGLE INC.

Inventor： Yifang Liu ,  Clemens Buehling ,  Fei Ye

IPC: G06F7/00 ,  G06F17/30 ,  G06F17/00

CPC classification number: G06F17/30073 ,  G06F17/3007 ,  G06F17/30091 ,  G06Q10/10 ,  G06Q50/01

Abstract: A system and method for social-aware clustering of user data replicas in a large-scale distributed computing system is disclosed. An exemplary system finds at least one user's connected users based on communications between the user and other users. The datacenters that contain the user replicas of the user's connected users are found. Connections and connection weights between the user and the user's connected users' datacenters are computed. The preferred datacenters for the user's current user data replica is computed based on the location of the connected datacenters and the weights of the connections. An optimization model minimizes the distance between the user's current datacenter and the user's preferred datacenter to reduce network traffic and central processing unit usage and determines the user's datacenter. The user's current datacenter is updated to the datacenter determined by running the optimization model.

公开(公告)号：US09767489B1

公开(公告)日：2017-09-19

申请号：US14032169

申请日：2013-09-19

Applicant: Google Inc.

Inventor： Yifang Liu ,  Konstantinos Katsiapis ,  Christopher Kenneth Harris

IPC: G06Q30/02 ,  H04L29/08 ,  G06F17/30

CPC classification number: G06Q30/0277 ,  G06F17/30867 ,  G06Q30/0241 ,  G06Q30/0249 ,  G06Q30/0253 ,  H04L67/22 ,  H04L67/32

Abstract: When a content item is initially served to a client device, the content item may result in an impression effect. As time elapses, the initial impression may fade. Such a decay of the impression effect may be predicted through the use of a predictive model. In some implementations, one or more impression effect parameters may be accessed and used with the predictive model to determine a decay factor or predicted value that incorporates the impression effect decay for a content item. A value, such as a score, may be determined based on the decay factor or the predicted value and a bid associated with a content item. A content item may be selected based on the determined value and data to effect presentation of the content item may be provided.

公开(公告)号：US20170236171A1

公开(公告)日：2017-08-17

申请号：US14449020

申请日：2014-07-31

Applicant: Google Inc.

Inventor： Yifang Liu

IPC: G06Q30/02

CPC classification number: G06Q30/0283 ,  G06Q30/0275

Abstract: Systems and methods for determining a relative pricing indication of content item criteria are provided. One method includes retrieving content item data relating to a plurality of content items. For each content item, a target pricing parameter and one or more selection criteria associated with the content item is determined, and the content item is categorized within one or more categories. For each of the categories, category pricing parameter data is generated based on a combination of the target pricing parameters for the content items within the category. For each of the selection criteria, a criteria pricing parameter is determined based on a combination of target pricing parameters for the content items with which the criterion is associated. Criteria pricing parameter data is correlated to the category pricing parameters for the one or more categories. Relative pricing indication data is generated for the criterion based on the correlation.

公开(公告)号：US09747346B1

公开(公告)日：2017-08-29

申请号：US14452952

申请日：2014-08-06

Applicant: Google Inc.

Inventor： Yifang Liu

IPC: G06F17/30 ,  G06Q30/02 ,  G06F15/16

CPC classification number: G06F17/3053 ,  G06F17/30061 ,  G06Q30/0259

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for providing content items based on a location of a user's attention in a map interface. In one aspect, a method includes receiving data specifying one or more map attention spots in a viewport of a map interface presented at a user device. Each map attention spot is a location on a map of the map interface corresponding to a specified amount of user activity. Content items are identified for presentation with the map. For each content item and map attention spot, a distance between a presentation location for the content item on the map and a location of the identified map attention spot on the map is determined. A rank score for the content item based at least on a respective content item score for the content item and each determined distance for the content item.

公开(公告)号：US09697474B2

公开(公告)日：2017-07-04

申请号：US14096050

申请日：2013-12-04

Applicant: Google Inc.

Inventor： Yifang Liu ,  Konstantinos Katsiapis ,  Samuel Ieong ,  Roberto Bayardo

IPC: G06F17/00 ,  G06N5/04 ,  G06N99/00

CPC classification number: G06N99/005

Abstract: Multi-class classification by training a machine learning system based on training inputs each of which includes features and at least one class label. Each training input is assigned a membership value that can indicate if an entity having the features of the training input is a member of the class corresponding to the class label that is also included in the training input. To determine if an entity having test features is a member of several test classes, test inputs can be constructed where each input includes the test features and a class label corresponding to one of the test classes. The test inputs are processed by the trained machine learning system, which produces as outputs test membership values that represent the likelihood that the entity having the features in the test input belong to the test class corresponding to the test class label also included in the test input.

公开(公告)号：US09600769B1

公开(公告)日：2017-03-21

申请号：US14099114

申请日：2013-12-06

Applicant: GOOGLE INC.

Inventor： Yifang Liu ,  Clemens Buehling

IPC: G06N5/02

CPC classification number: G06N5/02 ,  G06N5/022 ,  H04L51/02 ,  H04L51/32

Abstract: Provided are methods and systems for constructing a personal knowledge graph for a user based on data contained in existing e-mail messages of the user, and using the personal knowledge graph to provide the user with contextually-relevant content and/or contact suggestions while the user is composing an e-mail message. A personal knowledge graph is constructed based on relations/connections between users and content identified from data contained in e-mail messages sent and/or received by the user. Such relations include content-content relations, user-content relations, and user-(content)-user relations. When a user is composing an e-mail message, the system responsively processes, analyzes, and indexes composing e-mail message data. The composing e-mail message data is used to fetch relevant information from the user's personal knowledge graph and generate one or more content and/or contact suggestions for presentation to the user alongside an e-mail message composing view.

公开(公告)号：US20150154507A1

公开(公告)日：2015-06-04

申请号：US14096050

申请日：2013-12-04

Applicant: Google, Inc.

Inventor： Yifang Liu ,  Konstantinos Katsiapis ,  Samuel Ieong ,  Roberto Bayardo

IPC: G06N99/00

CPC classification number: G06N99/005

Abstract: Multi-class classification by training a machine learning system based on training inputs each of which includes features and at least one class label. Each training input is assigned a membership value that can indicate if an entity having the features of the training input is a member of the class corresponding to the class label that is also included in the training input. To determine if an entity having test features is a member of several test classes, test inputs can be constructed where each input includes the test features and a class label corresponding to one of the test classes. The test inputs are processed by the trained machine learning system, which produces as outputs test membership values that represent the likelihood that the entity having the features in the test input belong to the test class corresponding to the test class label also included in the test input.

Abstract translation: 通过训练基于训练输入的机器学习系统进行多类分类，每个训练输入包括特征和至少一个类标签。 每个训练输入被分配成员值，该值可以指示具有训练输入特征的实体是否也包括在训练输入中的类标签对应的类的成员。 为了确定具有测试特征的实体是否是几个测试类的成员，可以构建测试输入，其中每个输入包括测试特征和与其中一个测试类相对应的类标签。 测试输入由经过训练的机器学习系统处理，其产生作为输出测试成员资格值的输出，该成员值表示具有测试输入中的特征的实体属于与测试类标签相对应的测试类别的可能性，该测试类也包括在测试输入中 。

公开(公告)号：US20140122697A1

公开(公告)日：2014-05-01

申请号：US13666391

申请日：2012-11-01

Applicant: Google Inc.

Inventor： Yifang Liu ,  Roberto J. Bayardo, JR. ,  Guangyu Zhu

IPC: G06F15/173

CPC classification number: G06F16/9535

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer-readable storage medium, and including a method for selecting content. The method comprises identifying historical data including a log of entries, each entry including a user identifier associated with a given user that accessed a service, a service identifier associated with the service, and a device identifier associated with a device used by the user for accessing the service. The method further comprises evaluating log entries including identifying pairs of log entries that include a same user identifier and a same service identifier but different device identifiers. The method further comprises linking plural devices to a user associated with the same user identifier based on the evaluating when a given device is determined to be likely to be personal to the user, and using historical information associated with the linked devices to select content for delivery to the user.

Abstract translation: 方法，系统和装置，包括在计算机可读存储介质上编码的计算机程序，并且包括用于选择内容的方法。 该方法包括识别包括条目日志的历史数据，每个条目包括与访问服务的给定用户相关联的用户标识符，与该服务相关联的服务标识符以及与用户访问的设备相关联的设备标识符 服务。 该方法还包括评估日志条目，包括识别包括相同用户标识符和相同服务标识符但不同设备标识符的日志条目对。 该方法还包括：基于当给定设备被确定为对用户可能是个人的评估时，将多个设备链接到与相同用户标识符相关联的用户，以及使用与所链接的设备相关联的历史信息来选择要传送的内容 给用户

公开(公告)号：US09727818B1

公开(公告)日：2017-08-08

申请号：US14187291

申请日：2014-02-23

Applicant: Google Inc.

Inventor： Yifang Liu

IPC: G06N5/02

CPC classification number: G06N5/02 ,  G06F17/30 ,  G06F17/30867 ,  G06Q30/00

Abstract: The modeling of an impression effect may include generating a content item impression effect distribution. A classification model may be used to determine a period of the content item impression effect distribution based on one or more accessed impression effect parameters. A value for a content item may be determined based, at least in part, on the determined period and a bid associated with the content item. A content item may be selected based on the determined value and data to display the selected content item may be transmitted. In some instances, the determined period may be used to determine or select predictive model for the determined period that outputs a factor to modify the determined value.