Symbolic Execution State Feature Selection for Software Testing

The patent extracts and selects only the most relevant state features from the complete state space using a selection strategy based on branch conditional statements and path conditional expressions. This extraction approach reduces the exponential state space to a manageable subset of critical features that maintain error detection capability while avoiding state explosion.

The patent transforms the state representation by changing parameters from complete state traces to condensed feature vectors derived from branch and path conditional expressions. This parameter transformation reduces dimensionality while preserving the essential information needed for effective software testing and error detection.

The patent enables the system to automatically generate and optimize state features and ranking functions using machine learning techniques, eliminating the need for expensive expert intervention. The system self-services by learning from execution data and automatically determining which branch conditional statements and path conditional expressions are most relevant for state selection.

The patent replaces the mechanical process of expert intuition and manual feature definition with automated machine learning algorithms. This substitution uses computational models to learn optimal state features from data, replacing human expert knowledge with an automated system that reduces development costs while maintaining or improving effectiveness.

The patent applies local quality by assigning different importance weights to different branch conditional statements based on their relevance to error detection. Instead of treating all features equally, the system identifies and emphasizes locally critical features (those most relevant to specific error conditions) while downweighting or eliminating less important features, thereby reducing dimensionality while maintaining coverage.

The patent implements partial action by selecting only the necessary subset of branch conditional statements that provide sufficient testing coverage. Rather than using all possible features, the system identifies the minimal adequate set of features needed to detect errors, reducing feature vector dimensionality while maintaining effective coverage through selective feature inclusion.