Distributed processing in a messaging platform

A system comprising:

• one or more computers and one or more storage devices on which are stored instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:
• receiving a graph (G) comprising a plurality of nodes that each represent a respective account of a messaging system and a plurality of edges that each represent a relationship between a pair of accounts, wherein the accounts represented by G comprise a plurality of targets and a plurality of influencers;
• value partitioning a transpose of the graph (GT) to generate a plurality of value partitioned graphs and distributing the value partitioned graphs across a plurality of physical shards according to a distribution scheme;
• key partitioning (G) to generate a plurality of key partitioned graphs and distributing the key partitioned graphs across the plurality of physical shards according to the distribution scheme;
• receiving an action graph that includes actions performed by influencers including an action performed by a first influencer;
• receiving a request to determine whether to provide a recommendation to a first target based on the action, wherein in response to the request querying each physical shard to perform actions comprising:
• looking-up value partitioned graphs of GT using the first influencer as key to obtain a set of targets, the set of targets including the first target;
• looking-up key partitioned graphs of G using each of the set of targets as key to obtain a set of influencers of the set of targets;
• and
• intersecting the set of influencers with the action graph for a specified timeframe to determine which user accounts of the set of influencers have performed the action within the timeframe;
• receiving query results from one or more physical shards, wherein the results from each shard indicate which influencers performed the action within the timeframe;
• using the received query results, determining whether to provide a recommendation to the first target based on the actions being performed within the timeframe by a specified number of influencers of the first target;
• and
• in response to a determination to send the recommendation to the first target, sending the recommendation to the first target.

A non-transitory computer readable medium comprising computer readable program code, which when executed by a computer processor causes the computer processor to perform operations comprising:

• receiving a graph (G) comprising a plurality of nodes that each represent a respective account of a messaging system and a plurality of edges that each represent a relationship between a pair of accounts, wherein the accounts represented by G comprise a plurality of targets and a plurality of influencers;
• value partitioning a transpose of the graph (GT) to generate a plurality of value partitioned graphs and distributing the value partitioned graphs across a plurality of physical shards according to a distribution scheme;
• key partitioning (G) to generate a plurality of key partitioned graphs and distributing the key partitioned graphs across the plurality of physical shards according to the distribution scheme;
• receiving an action graph that includes actions performed by influencers including an action performed by a first influencer;
• receiving a request to determine whether to provide a recommendation to a first target based on the action, wherein in response to the request querying each physical shard to perform actions comprising:
• looking-up value partitioned graphs of GT using the first influencer as key to obtain a set of targets, the set of targets including the first target;
• looking-up key partitioned graphs of G using each of the set of targets as key to obtain a set of influencers of the set of targets;
• and
• intersecting the set of influencers with the action graph for a specified timeframe to determine which user accounts of the set of influencers have performed the action within the timeframe;
• receiving query results from one or more physical shards, wherein the results from each shard indicate which influencers performed the action within the timeframe;
• using the received query results, determining whether to provide a recommendation to the first target based on the actions being performed within the timeframe by a specified number of influencers of the first target;
• and
• in response to a determination to send the recommendation to the first target, sending the recommendation to the first target.