Data Obsolescence Detection in the Light of Newly Acquired Valid Observations.

摘要: The information describing the conditions of a system or person is constantly evolving and may become obsolete contradict other information. A database, therefore, must be consistently updated upon acquisition new valid observations that ones contained in database. In this paper, we propose novel approach for dealing with obsolescence problem. Our aims to detect, real-time, contradictions between then identify ones, given representation model. Since work within an uncertain environment characterized by lack information, choose use Bayesian network as our model approximate concept, $\epsilon$-Contradiction. concept parameterised confidence level having contradiction set observations. We polynomial-time algorithm detecting show resulting better represented AND-OR tree than simple Finally, demonstrate effectiveness on real elderly fall-prevention database showcase how can used give reliable recommendations doctors. experiments systematically substantially very good results.