公开(公告)号：US09449271B2

公开(公告)日：2016-09-20

申请号：US14834274

申请日：2015-08-24

Applicant: Google Inc.

Inventor： Qingzhou Wang ,  Yu Liang ,  Ke Yang ,  Kai Chen

IPC: G06F15/18 ,  G06E1/00 ,  G06E3/00 ,  G06G7/00 ,  G06N3/04 ,  G06N3/02 ,  G06F17/30 ,  G06N3/08 ,  G06K9/62 ,  G06F3/0484 ,  G06N7/00

CPC classification number: G06N3/04 ,  G06F3/0484 ,  G06F17/3053 ,  G06F17/30707 ,  G06F17/30864 ,  G06K9/627 ,  G06N3/02 ,  G06N3/0427 ,  G06N3/084 ,  G06N7/005

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scoring concept terms using a deep network. One of the methods includes receiving an input comprising a plurality of features of a resource, wherein each feature is a value of a respective attribute of the resource; processing each of the features using a respective embedding function to generate one or more numeric values; processing the numeric values using one or more neural network layers to generate an alternative representation of the features, wherein processing the floating point values comprises applying one or more non-linear transformations to the floating point values; and processing the alternative representation of the input using a classifier to generate a respective category score for each category in a pre-determined set of categories, wherein each of the respective category scores measure a predicted likelihood that the resource belongs to the corresponding category.

公开(公告)号：US20170061025A1

公开(公告)日：2017-03-02

申请号：US15350377

申请日：2016-11-14

Applicant: Google Inc.

Inventor： Neha Arora ,  Ke Yang ,  Zuguang Yang

IPC: G06F17/30

CPC classification number: G06F17/30867 ,  G06F17/30241 ,  G06F17/30424 ,  G06F17/3053 ,  G06F17/30554 ,  G06F17/30864

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for enforcing the category diversity or sub-category diversity of POIs that are identified in response to a local search. According to one implementation, a method includes receiving a request to identify points of interest (POIs), obtaining data identifying (i) candidate points of interest (POIs) that satisfy the request, (ii) a respective category associated with each candidate POI, and (iii) a non-scaled score associated with each candidate POI, and ranking, for each of one or more of the categories, the candidate POIs associated with the category, based on the respective non-scaled scores. The method also includes scaling, for each of the one or more categories, the non-scaled scores of the ranked candidate POIs associated with the category, ranking the candidate POIs using the scaled scores, for the candidate POIs that are associated with the one or more categories, and the non-scaled scores, for the candidate POIs that are not associated with the one or more categories, and providing data that identifies two or more of the candidate POIs, as ranked according to the scaled scores and the non-scaled scores.

Abstract translation: 方法，系统和装置，包括在计算机存储介质上编码的计算机程序，用于执行响应于本地搜索而识别的POI的类别分集或子类别多样性。 根据一个实施方式，一种方法包括接收识别兴趣点（POI）的请求，获得识别（i）满足该请求的候选兴趣点（POI）的数据，（ii）与每个候选POI相关联的相应类别， 以及（iii）与每个候选POI相关联的非缩放分数，并且基于相应的非缩放分数，针对所述类别中的一个或多个类别中的每一个对与该类别相关联的候选POI进行排名。 该方法还包括对于所述一个或多个类别中的每一个，对与该类别相关联的排名候选POI的未缩放分数进行缩放，使用缩放分数对与候选POI相关联的候选POI进行排序， 更多类别以及与该一个或多个类别不相关联的候选POI的非缩放分数，以及提供标识候选POI中的两个或更多个的数据，如根据缩放分数和未缩放分数 分数。

公开(公告)号：US20140279774A1

公开(公告)日：2014-09-18

申请号：US13802462

申请日：2013-03-13

Applicant: Google Inc.

Inventor： Qingzhou Wang ,  Yu Liang ,  Ke Yang ,  Kai Chen

IPC: G06N3/02

CPC classification number: G06N3/04 ,  G06F3/0484 ,  G06F17/3053 ,  G06F17/30707 ,  G06F17/30864 ,  G06K9/627 ,  G06N3/02 ,  G06N3/0427 ,  G06N3/084 ,  G06N7/005

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scoring concept terms using a deep network. One of the methods includes receiving an input comprising a plurality of features of a resource, wherein each feature is a value of a respective attribute of the resource; processing each of the features using a respective embedding function to generate one or more numeric values; processing the numeric values using one or more neural network layers to generate an alternative representation of the features, wherein processing the floating point values comprises applying one or more non-linear transformations to the floating point values; and processing the alternative representation of the input using a classifier to generate a respective category score for each category in a pre-determined set of categories, wherein each of the respective category scores measure a predicted likelihood that the resource belongs to the corresponding category.

Abstract translation: 方法，系统和装置，包括在计算机存储介质上编码的计算机程序，用于使用深层网络评分概念术语。 所述方法之一包括接收包括资源的多个特征的输入，其中每个特征是所述资源的相应属性的值; 使用相应的嵌入功能处理每个特征以生成一个或多个数值; 使用一个或多个神经网络层处理所述数值以产生所述特征的替代表示，其中处理所述浮点值包括对所述浮点值应用一个或多个非线性变换; 以及使用分类器处理所述输入的替代表示以针对预定类别集合中的每个类别生成相应的类别分数，其中各个类别分数中的每一个测量所述资源属于相应类别的预测可能性。

公开(公告)号：US20160048754A1

公开(公告)日：2016-02-18

申请号：US14834274

申请日：2015-08-24

Applicant: Google Inc.

Inventor： Qingzhou Wang ,  Yu Liang ,  Ke Yang ,  Kai Chen

IPC: G06N3/04 ,  G06F17/30 ,  G06N7/00 ,  G06F3/0484

CPC classification number: G06N3/04 ,  G06F3/0484 ,  G06F17/3053 ,  G06F17/30707 ,  G06F17/30864 ,  G06K9/627 ,  G06N3/02 ,  G06N3/0427 ,  G06N3/084 ,  G06N7/005

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scoring concept terms using a deep network. One of the methods includes receiving an input comprising a plurality of features of a resource, wherein each feature is a value of a respective attribute of the resource; processing each of the features using a respective embedding function to generate one or more numeric values; processing the numeric values using one or more neural network layers to generate an alternative representation of the features, wherein processing the floating point values comprises applying one or more non-linear transformations to the floating point values; and processing the alternative representation of the input using a classifier to generate a respective category score for each category in a pre-determined set of categories, wherein each of the respective category scores measure a predicted likelihood that the resource belongs to the corresponding category.

Abstract translation: 方法，系统和装置，包括在计算机存储介质上编码的计算机程序，用于使用深层网络评分概念术语。 所述方法之一包括接收包括资源的多个特征的输入，其中每个特征是所述资源的相应属性的值; 使用相应的嵌入功能处理每个特征以生成一个或多个数值; 使用一个或多个神经网络层处理所述数值以产生所述特征的替代表示，其中处理所述浮点值包括对所述浮点值应用一个或多个非线性变换; 以及使用分类器处理所述输入的替代表示以针对预定类别集合中的每个类别生成相应的类别分数，其中各个类别分数中的每一个测量所述资源属于相应类别的预测可能性。

公开(公告)号：US09147154B2

公开(公告)日：2015-09-29

申请号：US13802462

申请日：2013-03-13

Applicant: Google Inc.

Inventor： Qingzhou Wang ,  Yu Liang ,  Ke Yang ,  Kai Chen

IPC: G06F15/18 ,  G06E1/00 ,  G06E3/00 ,  G06G7/00 ,  G06N3/02 ,  G06F17/30 ,  G06N3/04 ,  G06N3/08 ,  G06K9/62

CPC classification number: G06N3/04 ,  G06F3/0484 ,  G06F17/3053 ,  G06F17/30707 ,  G06F17/30864 ,  G06K9/627 ,  G06N3/02 ,  G06N3/0427 ,  G06N3/084 ,  G06N7/005

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scoring concept terms using a deep network. One of the methods includes receiving an input comprising a plurality of features of a resource, wherein each feature is a value of a respective attribute of the resource; processing each of the features using a respective embedding function to generate one or more numeric values; processing the numeric values using one or more neural network layers to generate an alternative representation of the features, wherein processing the floating point values comprises applying one or more non-linear transformations to the floating point values; and processing the alternative representation of the input using a classifier to generate a respective category score for each category in a pre-determined set of categories, wherein each of the respective category scores measure a predicted likelihood that the resource belongs to the corresponding category.

Abstract translation: 方法，系统和装置，包括在计算机存储介质上编码的计算机程序，用于使用深层网络评分概念术语。 所述方法之一包括接收包括资源的多个特征的输入，其中每个特征是所述资源的相应属性的值; 使用相应的嵌入功能处理每个特征以生成一个或多个数值; 使用一个或多个神经网络层处理所述数值以产生所述特征的替代表示，其中处理所述浮点值包括对所述浮点值应用一个或多个非线性变换; 以及使用分类器处理所述输入的替代表示以针对预定类别集合中的每个类别生成相应的类别分数，其中各个类别分数中的每一个测量所述资源属于相应类别的预测可能性。

公开(公告)号：US08990210B2

公开(公告)日：2015-03-24

申请号：US13968339

申请日：2013-08-15

Applicant: Google Inc.

Inventor： Daniel Egnor ,  Paul Haahr ,  Kevin D. Lacker ,  John Ogden Lamping ,  Amitabh K. Singhal ,  Ke Yang

IPC: G06F7/00 ,  G06F17/30

CPC classification number: G06F17/30321 ,  G06F17/30864 ,  G06F17/3089

Abstract: Web pages of a Website may be processed to improve search results. For example, information likely to pertain to more than just the Web page it is directly associated with may be identified. One or more other, related, Web pages that such information is likely to pertain to is also identified. The identified information is associated with the identified other Web page(s) and this association is saved in a way to affect a search result score of the Web page(s).

Abstract translation: 可能会处理网站的网页以改进搜索结果。 例如，可能会识别与其直接相关联的网页不仅仅在于其信息的信息。 也可以识别出这样的信息可能涉及的一个或多个其他相关的网页。 所识别的信息与所标识的其他网页相关联，并且以影响网页的搜索结果分数的方式保存该关联。