Towards sequenced semantics for evolving graphs

The research community has adopted a sequence of snapshots as the logical representation of evolving graphs — graphs that change over time and whose history of evolution we want to preserve for analysis. This paper argues that the snapshot sequence model of evolving graphs is insufficient for representation and analysis of a wide range of networks. Instead, we propose to use the interval model with sequenced semantics. In this model nodes and edges are associated with their validity intervals, and operations adhere to the properties of snapshot reducibility, extended snapshot reducibility, and change preservation. We show the advantages of adopting this model for evolving graphs and lay the groundwork for an evolving graph query language with sequenced semantics. We also discuss several challenges of efficiently supporting sequenced semantics in a distributed setting.