Enhancing context specifications for dependable adaptive systems: A data mining approach

摘要: Abstract Context: Adaptive systems are expected to cater for various operational contexts by having multiple strategies in achieving their objectives and the logic matching an actual context. The prediction of relevant at design time is paramount dependability. With current trend on using data mining support requirements engineering process, this task understanding context adaptive system can benefit from such techniques as well.Objective: objective provide a method refine specification contextual variables relation This refinement shall detect dependencies between variables, priorities monitoring them, decide relevance choosing right strategy decision tree.Method: Our requirements-driven approach adopts goal modelling structure addition operationalization values sensed information map system’s behaviour. We propose analysis process subset algorithms extract list related tasks, and/or goals.Results: experimentally evaluated our proposal Body Sensor Network (BSN), simulating 12 resources that could lead variability space 4096 possible conditions. was able elicit subtle would significantly affect service provided assisted patients relations contexts, assisting need, priority monitoring.Conclusion: use some mitigate lack precise definition practical supportive traditional methods, which typically require intense human intervention.