Method and medium for implicit relationship discovery based on cumulative co-temporal activity

Abstract

Implied relationships between entities, such as network endpoints, are automatically discovered based on co-temporal events. Events involving pairs of endpoints, such as messaging events in which one endpoint acts as a source and another endpoint acts as a destination, may be detected. Edges between nodes representing those endpoints and other nodes representing other endpoints involved in other recent (co-temporal) events may be added to a progressively constructed graph. Over time, such edges may be progressively weighted in response to the detection of further co-temporal events involving the same endpoints. Relationships between endpoints may be implied based on the resulting accumulated weights of edges linking those endpoints' nodes in the graph even if there is no express relationship between those endpoints in any real-word context (e.g., even if those endpoints are not directly connected in any network, and even if no single event involves both of those endpoints together).