Incorporating quality aspects in sensor data streams

  • Authors:
  • Anja Klein

  • Affiliations:
  • SAP AG, Dresden, Germany

  • Venue:
  • Proceedings of the ACM first Ph.D. workshop in CIKM
  • Year:
  • 2007

Quantified Score

Hi-index 0.00

Visualization

Abstract

Sensors are increasingly embedded into physical products in order to capture data about their conditions and usage for decision making in business applications. However, a major issue for such applications is the limited quality of the captured data due to inherently restricted precision and performance of the sensors. Moreover, the data quality is further decreased by data processing to meet resource constraints in streaming environments and ultimately influences business decisions. The issue of how to efficiently provide applications with information about data quality (DQ) is still an open research problem. In my Ph.D. thesis, I address this problem by developing a system to provide business applications with accurate information on data quality. Furthermore, the system will be able to incorporate and guarantee user-defined data quality levels. In this paper, I will present the major results from my research so far. This includes a novel jumping-window-based approach for the efficient transfer of data quality information as well as a flexible metamodel for storage and propagation of data quality. The comprehensive analysis of common data processing operators w.r.t. their impact on data quality allows a fruitful knowledge evaluation and thus diminishes incorrect business decisions.