Unifying Graph Traversals and Time Series Joins in Hybrid Graphs

Graphs are highly expressive data structures for modelling and analysing relationships between real-world objects, with several applications, such as fraud detection, social and transportation networks. Traditionally, graphs and time series have been treated and studied as separate entities. However, the growing complexity of dynamic systems has highlighted the need to unify and analyse them altogether. In this paper, we address this gap for the first time by proposing an integrated approach that combines graph topology with time series data, enabling a new class of analytical queries that capture both evolving relationships and temporal patterns. We present the first declarative subgraph matching algorithm for hybrid graphs, where we leverage the subsequence join operation enriched with Allen's interval algebra to navigate the graph by computing the recurrent patterns present in time series: in this way, we can traverse the graph employing topological data and time series data to construct the resulting path. Our extensive experimental study demonstrates the efficiency and scalability of integrating time series similarity with graph pattern analysis.