公开(公告)号：US20170185601A1

公开(公告)日：2017-06-29

申请号：US14983321

申请日：2015-12-29

Applicant: Facebook, Inc.

Inventor： Yang Qin ,  Zhong Zhang ,  Sergiu-Cosmin Ferentz ,  Dan Ionut Fechete

IPC: G06F17/30

CPC classification number: G06F16/24578 ,  G06F16/9535 ,  G06N5/022 ,  G06N7/005 ,  G06Q50/01

Abstract: In one embodiment, a method includes, in an online social network, accessing one or more first content objects associated with a user in the online social network and a second content object, determining topics and corresponding first weights of the topics for the first content objects using a topic extraction algorithm, where each first weight indicates a strength of an association between the corresponding topic and the first content object(s), determining one or more second weights of the topics for the second content object, where each second weight indicates a strength of an association between the corresponding topic and the second content object, and calculating a similarity score for the second content object based on a comparison of, for each topic, the first weight to the second weight, where the similarity score is to be used in a determination regarding presentation of the second content object to the user.

公开(公告)号：US10979870B1

公开(公告)日：2021-04-13

申请号：US15961789

申请日：2018-04-24

Applicant: Facebook, Inc.

Inventor： Zhong Zhang

IPC: G06Q10/10 ,  G06Q10/08 ,  G06Q30/06 ,  G06Q30/02 ,  G06Q10/06 ,  H04W4/18 ,  G06Q50/00 ,  H04W4/029 ,  G06F16/29

Abstract: In one embodiment, a method includes, by one or more computing devices, receiving, from a client system of a user, a request for event recommendations for the user; accessing information indicating a geographic location associated with the user; accessing a geographic map comprising multiple map tiles, each map tile defining a geographic area within the map; identifying a first map tile of the multiple map tiles associated with the user based on the geographic location associated with the user; and determining a social tile associated with the first map tile, wherein the social tile comprises the first map tile and one or more second map tiles of the multiple map tiles, wherein the first map tile and the one or more second map tiles are clustered into the social tile based on one or more items of prior attendee information corresponding to one or more prior events.

公开(公告)号：US10409823B2

公开(公告)日：2019-09-10

申请号：US14983321

申请日：2015-12-29

Applicant: Facebook, Inc.

Inventor： Yang Qin ,  Zhong Zhang ,  Sergiu-Cosmin Ferentz ,  Dan Ionut Fechete

IPC: G06F16/2457 ,  G06F16/9535 ,  G06Q50/00

Abstract: In one embodiment, a method includes, in an online social network, accessing one or more first content objects associated with a user in the online social network and a second content object, determining topics and corresponding first weights of the topics for the first content objects using a topic extraction algorithm, where each first weight indicates a strength of an association between the corresponding topic and the first content object(s), determining one or more second weights of the topics for the second content object, where each second weight indicates a strength of an association between the corresponding topic and the second content object, and calculating a similarity score for the second content object based on a comparison of, for each topic, the first weight to the second weight, where the similarity score is to be used in a determination regarding presentation of the second content object to the user.

公开(公告)号：US20170034108A1

公开(公告)日：2017-02-02

申请号：US14814382

申请日：2015-07-30

Applicant: Facebook, Inc.

Inventor： Dan Ionut Fechete ,  Jun Jiang ,  Yang Qin ,  Zhong Zhang

IPC: H04L12/58 ,  H04L29/08

CPC classification number: H04L51/32 ,  G06N5/022 ,  G06N20/00 ,  G06Q30/02 ,  H04L51/14 ,  H04L67/02 ,  H04L67/20 ,  H04L67/306

Abstract: In one embodiment, a method includes, by a computing device, identifying an event in an online social network to be evaluated for recommendation to a user of the online social network and determining whether the event is recommendable to the user, the determination being based on identifying correlations between one or more characteristics of the user and a plurality of signals associated with the event. The method further includes, in response to determining that the event is recommendable, presenting a recommendation or promotion for the event to the user, and, in response to determining that the event is not recommendable, converting the event in accordance with the determining that the event is not recommendable. The signals may include content associated with the event, metadata associated with the event, or responses to a notification about the event by users of the online social network.

Abstract translation: 在一个实施例中，一种方法包括通过计算设备识别要在线社交网络中被评估以推荐给在线社交网络的用户的事件并且确定该事件是否对用户可推荐的，所述确定基于 识别用户的一个或多个特征与与该事件相关联的多个信号之间的相关性。 所述方法还包括响应于确定所述事件是可推荐的，向所述用户呈现所述事件的推荐或促销，并且响应于确定所述事件不可推荐，根据确定所述事件 事件不推荐 信号可以包括与事件相关联的内容，与事件相关联的元数据，或者对在线社交网络的用户对关于事件的通知的响应。

公开(公告)号：US11080349B1

公开(公告)日：2021-08-03

申请号：US15981724

申请日：2018-05-16

Applicant: Facebook, Inc.

Inventor： Zhong Zhang ,  Jin Fang

IPC: G06N5/04 ,  G06F16/9535 ,  G06N20/00

Abstract: In one embodiment, a method includes generating embeddings for social-networking entities by training the embeddings using a training algorithm, where an embedding corresponding to an entity represents a point in a d-dimensional embedding space, identifying a subset of entities having one or more common attributes that is not encoded in the generated embeddings, encoding, for each entity in the subset, values of the one or more common attributes into a j-dimensional additional embedding, creating, for each entity in the subset, a (d+j)-dimensional embedding by concatenating the generated d-dimensional embedding with the j-dimensional additional embedding, detecting a need to identify entities similar to a reference entity that is a member of the subset, computing k-nearest neighbors of an embedding corresponding to the reference entity in the (d+j)-dimensional embedding space, identifying entities corresponding to the computed k-nearest neighbors, and providing information regarding the corresponding entities.