公开(公告)号：US11176210B1

公开(公告)日：2021-11-16

申请号：US16128142

申请日：2018-09-11

Applicant: Amazon Technologies, Inc.

Inventor： John Martin Miller ,  Anuj Chadha ,  Anthony Crider

IPC: G06F16/00 ,  G06F16/9535 ,  G06F16/35 ,  G06F16/955

Abstract: Techniques for improving dynamic content retrieval are described. In an example, a system may receive a request for content. The request may be associated with an access by a computing device to a network document. The system may determine a context associated with the network document based on the request. The content may be pre-computed based on historical accesses to a network resource. The system may determine a match between the context and a pre-computed context that is associated with content and may select the content from a set of candidate content based on an intent description of a content provider and historical behavior data of visitors to the network resource. The intent description may be associated with a presentation of the content at the network resource. The system may provide an identifier of the content. The identifier may be usable to present the content at the computing device.

公开(公告)号：US11023554B1

公开(公告)日：2021-06-01

申请号：US16128073

申请日：2018-09-11

Applicant: Amazon Technologies, Inc.

Inventor： John Martin Miller ,  Anuj Chadha ,  Anthony Crider

IPC: G06F16/00 ,  G06F16/957 ,  G06F3/0482 ,  G06F9/451 ,  G06F16/955 ,  G06F16/958 ,  G06F16/9535 ,  G06F40/279

Abstract: Techniques for improving dynamic content retrieval are described. In an example, a computer system may receive, from a content provider device, an intent description associated with providing content at a network resource. The computer system may determine a goal and a predicate based on the intent description and may select a workflow from a plurality of workflows based on the goal and the predicate. The workflow may be associated with determining a context of a network document based on historical behavior data of visitors to the network resource and on a taxonomy of the network resource. The computer system may determine the context based on an execution of the workflow, the historical behavior data, and the taxonomy. The computer system may store an association between the context and the content. The association may be usable to select the content for presentation in a network document of the network resource.