Modern applications (e.g., the so called Future Internet applications) exhibit properties that make them hard to model once for all. In fact, they dynamically adapt to the user's habits, to the context, to the environment; they dynamically discover new services and components to integrate; they modify themselves through reflection, autonomically. Model inference techniques are based on the observation of the application behavior (trace collection) and on its generalization into a model. Model inference supports testing, understanding and evolution of the software. However, inferred models may become obsolete at run time, due to the self modifications of the software. We investigate an approach for the automated detection of model discontinuities, based on a trade off between delay of the detection and accuracy, measured in terms of few false negatives.
Automated Detection of Discontinuities in Models Inferred from Execution Traces
Marchetto, Alessandro;Nguyen, Duy Cu;Tonella, Paolo
2011-01-01
Abstract
Modern applications (e.g., the so called Future Internet applications) exhibit properties that make them hard to model once for all. In fact, they dynamically adapt to the user's habits, to the context, to the environment; they dynamically discover new services and components to integrate; they modify themselves through reflection, autonomically. Model inference techniques are based on the observation of the application behavior (trace collection) and on its generalization into a model. Model inference supports testing, understanding and evolution of the software. However, inferred models may become obsolete at run time, due to the self modifications of the software. We investigate an approach for the automated detection of model discontinuities, based on a trade off between delay of the detection and accuracy, measured in terms of few false negatives.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
