Spatio-temporal template discovery using rough set theory

Abstract

Real-time stream data is characterized by spatial and temporal variability and is subject to unbounded or constantly evolving entities. The challenge is how to aggregate these unbounded data streams at different spaces and times to provide effective decisions making in real-time. This paper proposes a rough set-based sliding window framework for stream data aggregation. Based on current data streams, it identifies interesting spatio-temporal patterns, and generates rough set If … Then decision rules. Proposed formalism has been tested on sea surface temperature data from NOAA's TAO/TRITON project. Such a pattern-based data aggregation scheme has the potential to significantly reduce data communications in decision making.