Abstract: A system determines a set of users satisfying a given criteria based on attribute values of the users. The system associates each value of an attribute with a representation of users having that attribute value. The representation of users may be an enumeration of users or a bit index. The system receives a request for determining a set of users satisfying a given criterion. The system identifies a logical operation based on the criterion and performs the logical operation to determine the set of users. The representation of the set of users may be distributed across a plurality of processors, allowing the system to perform the logical operation in parallel. The system may further determine statistical information describing a characteristic of the set of users, for example, page likes by the set of users. The statistical information may be presented to a business for analyzing their target audience.

Abstract: Systems and methods for performing near real-time merging of distributed data streams are described. For example, streams of ad impressions, ad clicks, and conversions are sorted by user id into virtual buckets. The buckets of data are distributed across multiple servers, so that each server can process their respective buckets of data independently. Each server uses synchronization logic to determine a running delay distribution of the data streams. Based on the delay distributions, merge processing of the streams is appropriately delayed to ensure that the ad impression and ad click stream information needed for correlating with the conversion stream information is likely to be available in real time.

Abstract: Systems and methods for performing near real-time merging of distributed data streams are described. For example, streams of ad impressions, ad clicks, and conversions are sorted by user id into virtual buckets. The buckets of data are distributed across multiple servers, so that each server can process their respective buckets of data independently. Each server uses synchronization logic to determine a running delay distribution of the data streams. Based on the delay distributions, merge processing of the streams is appropriately delayed to ensure that the ad impression and ad click stream information needed for correlating with the conversion stream information is likely to be available in real time.

Abstract: A system determines a set of users satisfying a given criteria based on attribute values of the users. The system associates each value of an attribute with a representation of users having that attribute value. The representation of users may be an enumeration of users or a bit index. The system receives a request for determining a set of users satisfying a given criterion. The system identifies a logical operation based on the criterion and performs the logical operation to determine the set of users. The representation of the set of users may be distributed across a plurality of processors, allowing the system to perform the logical operation in parallel. The system may further determine statistical information describing a characteristic of the set of users, for example, page likes by the set of users. The statistical information may be presented to a business for analyzing their target audience.

Abstract: A social networking system receives a selection of user characteristics defining a benchmark audience and a target audience, and generates audience metrics that compare the audiences across a set of user characteristics. These user characteristics include demographics, interests, purchasing activity, and actions on the social networking system. The audience metrics are provided to an advertiser who may select additional user characteristics to refine the benchmark or target audiences. The audience metrics may include an affinity score that compares the audience metrics for a particular type of interaction, and may normalize the frequency of interactions relative to interactions of the audience as a whole. Advertisers may use the defined audiences to establish targeting criteria for an advertisement, and may use existing targeting criteria to seed the selection of an audience.

Abstract: A social networking system receives a selection of user characteristics defining a benchmark audience and a target audience, and generates audience metrics that compare the audiences across a set of user characteristics. These user characteristics include demographics, interests, purchasing activity, and actions on the social networking system. The audience metrics are provided to an advertiser who may select additional user characteristics to refine the benchmark or target audiences. The audience metrics may include an affinity score that compares the audience metrics for a particular type of interaction, and may normalize the frequency of interactions relative to interactions of the audience as a whole. Advertisers may use the defined audiences to establish targeting criteria for an advertisement, and may use existing targeting criteria to seed the selection of an audience.

Abstract: Systems and methods for performing near real-time merging of distributed data streams are described. For example, streams of ad impressions, ad clicks, and conversions are sorted by user id into virtual buckets. The buckets of data are distributed across multiple servers, so that each server can process their respective buckets of data independently. Each server uses synchronization logic to determine a running delay distribution of the data streams. Based on the delay distributions, merge processing of the streams is appropriately delayed to ensure that the ad impression and ad click stream information needed for correlating with the conversion stream information is likely to be available in real time.