Cleaning Uncertain Streams for Query Improvement

Real-world applications confront uncertain streams derived from unreliable data acquisition equipments and/or defective processing algorithms. However, application context covers specific cleaning rules to bring data close to the reality (i.e. data quality), and query features can filter data for the efficiency (i.e. data volume). in this paper, we propose a framework for cleaning uncertain data for query effectiveness and efficiency, which processes high-volume streams in parallel, and append new cleaning rules & queries seamlessly. We implement a prototype for video surveillance application over the architecture.