Khronos

Abstract

A temporal graph is a graph where its graph elements are only valid at a set of specific time. A concept of temporal graph traversals, which indicates the process of visiting temporally valid graph elements (i.e., events), is recently addressed. This kind of traversals can be used for analyzing information diffusion over time such as object traceability. However, it is possible that existing graph systems, not designed for temporal graph traversals, yield inefficiency in analyzing real-world temporal graphs. In the dissertation, we propose a temporal graph traversal platform, which enables efficient information diffusion analysis over time, called Khronos. Khronos provides the essential temporal syntax for property graph model and traversal language. In addition, Khronos manages indexes to efficiently traverse inter-events. In the evaluation, we show that the temporal syntax and indexes provided in the platform enhance the efficiency and outperforms existing property graph systems in terms of temporal graph traversals. As a case study, we apply Khronos to information systems for object traceability in order to show its feasibility. Our evaluation shows that applications can perform ad-hoc object traceability queries with less interaction and data transmission with the information systems, thus enhancing the scalability of the systems.