Generating metrics for content items presented in an online system

摘要

An online system allows creators and owners of content items presented within the system to view real-time metrics related to the content items. The online system stores data describing user interactions with content items in one or more activity logs and processes the activity logs in a distributed manner activity to identify user interactions with content items. For example, activity logs are divided into several portions that are processed in parallel using aggregation operators to generate user interaction aggregations. An aggregation includes one or more user interactions associated with the same content item and sharing certain criteria. From the user interaction aggregations, the online system determines one or more requested metrics, and may combine user interaction aggregations related to different content items when computing metrics.

主权项

1. A method comprising:
storing, in an online system, data describing interactions between one or more users of the online system and one or more content items presented by the online system; generating aggregations of types of interactions between one or more users and each of the one or more content items, wherein an aggregation of a type of interaction identifies one or more users performing the type of interaction on a content item; receiving a request for a metric describing interactions with one or more content items identified by the request and presented by the online system during a time interval; dividing stored interactions between one or more users of the online system and one or more of the content items identified by the request into a plurality of portions, the stored interactions not included in any of the aggregations; generating one or more intermediate aggregations of user interactions from the plurality of portions by applying an aggregation operator to user interactions included in the plurality of portions; and combining the one or more intermediate aggregations into a combined aggregation; retrieving one or more aggregations of types of interactions identifying one or more content items identified by the request; generating a value for the metric based on the combined aggregation and the retrieved one or more aggregations; and storing the value for the metric.