公开(公告)号：US20150227628A1

公开(公告)日：2015-08-13

申请号：US14175667

申请日：2014-02-07

Applicant: Yahoo! Inc.

Inventor： Supreeth Rao ,  Sunil Kumar Gupta ,  Chetan Nadgire

IPC: G06F17/30

CPC classification number: G06F17/30867

Abstract: A system can provide unique user metrics associated with a node in a hierarchy and the node's decedents in real time. This can occur by merging sketches of the node and its decedents to form a merged sketch, and retrieving the unique user metrics from the merged sketch. The sketches and the merged sketch may each be a separate data form, such as a hash. The hierarchy may include an online content taxonomy and the sketches of each node may include unique user metrics associated with the node, such as user interaction metrics associated with a category of content of the node. The system may build sketches for the nodes of a hierarchy, associate each sketch with its ancestor sketches, invert the associations, tie corresponding descendant sketches of an ancestor sketch based on the inversions, and merge the descendant sketches with their ancestor sketches to generate the merged sketches.

Abstract translation: 系统可以提供与层次结构中的节点相关联的唯一用户度量，并且可以实时地提供节点的节点。 这可以通过合并节点及其成员的草图来形成合并草图，并从合并草图中检索唯一用户度量。 草图和合并草图可以各自是单独的数据形式，例如散列。 层级可以包括在线内容分类法，并且每个节点的草图可以包括与节点相关联的唯一用户度量，诸如与节点的内容类别相关联的用户交互度量。 系统可以为层次结构的节点构建草图，将每个草图与其祖先草图相关联，反转关联，根据反转结合祖先草图的相应后代草图，并将后代草图与其祖先草图合并以生成合并 草图。