Event Association Analysis Using Graph Rules

Pengyue Jiang, Wenjun Wang, Xueli Liu, Lifei Sun, Bowen Dong

Abstract: Event association analysis is used to mine potential association relationships between different events in event data sets, and is widely used in areas such as commodity trading and social media networks. Graph rules can be defined with different forms and expressive capabilities according to relationships in various domains. For the event association domain, we define an event association rule based on statistical knowledge to examine positive and negative associations between events, and the generated patterns fuse other semantic information such as time and location to realize the inference on event data. We innovatively propose a matching method for matching candidate sets, which can further refine the rule matching results to each specific event node so that the results can be logically interpreted. Meanwhile, based on the existing rule matching algorithms, we propose an incremental computation method that can quickly process the incremental part of the event data, effectively saving computational resources and time. We demonstrate the accuracy and efficiency of the algorithm using real-life datasets.

External IDs:dblp:conf/icann/JiangWLSD23