Supporting Dynamic Graphs and Temporal Entity Deletions in the LDBC Social Network Benchmark's Data Generator

Many data processing pipelines operate on highly-connected data sets that can be efficiently modelled as graphs. These graphs are rarely static, but rather change rapidly and often exhibit dynamic, temporal, or streaming behaviour. During the last decade, numerous graph benchmarks have been proposed, which cover a significant portion of the features required in practical use cases. However, whilst these benchmarks often contain some update operations, none of them include complex deletions, which makes it challenging to test the performance of graph processing systems under such operations. To address this limitation, we have extended the LDBC Social Network Benchmark (SNB) by introducing lifespan attributes for the creation and deletion dates of its entities. We have defined constraints for selecting these dates from intervals that ensure that the graph always satisfies the cardinality constraints prescribed by the schema and other semantic constraints of the social network domain. We have implemented the proposed lifespans in the SNB generator.