Abstract
In this paper, we present a framework for managing qualitative spatiotemporal patterns. Our framework is designed for large scale monitoring systems. Such systems generate a huge amount of real-time data in various formats. End-users are interested in finding significant data configurations based on their expertise and attempt to leverage the large amounts of data generated by acquisition systems. Several software tools have been proposed to help users achieve such goals. However, available solutions are mostly based on relational databases and use SQL queries to support such functionalities. These systems do not allow for real-time detection of situations of interest also called 'patterns' in this domain due to the weak expressiveness of SQL queries. We present a novel approach based on complex event processing for the real-time detection of situations of interest based on events, states and spatial objects. We leverage the rich semantics of a qualitative pattern representation model to present a complete solution to qualitatively represent patterns and to detect their instances from the event cloud. Thanks to our approach, a user will be able to react to detected patterns instead of trying to identify 'to mine' patterns in databases as it is proposed in current approaches.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have