A parallel snapshot retrieval technique for MG

Abstract

©2016 ACM. Retrieving historical information in a short time is a crucial objective in dynamic graphs. Some of the existing dynamic graph models use log-files to store the graph evolution history. These log-files store the history as a set of events sorted by their occurrence time. For answering a historical query on these models, the snapshot at the given historic time instant must be constructed first. Since the log-file events are sorted by time, the construction of a historical snapshot mandates that all events must be parsed and applied till the target time. Thus, the snapshot construction consumes a lot of time. However, these models provide accurate results. That is why, it is of a significant importance to provide accurate and fast results. In this paper, we present a log-file structure that is able to efficiently handle both accuracy and performance.