원문정보
초록
영어
Path-oriented test data generation is in essence a Constraint Satisfaction Problem solved by search strategies, among which backtracking algorithms are widely used. In this paper, the backtracking algorithm Branch & Bound is introduced to generate path-oriented test data automatically. A model based on state space search is proposed to construct the search tree dynamically. Aiming at the programs containing constraints of strongly related variables even equalities, the static analysis technique interval arithmetic is optimized for the precise judgment of the assignment to each variable. The analysis on conflict is made accurate via distance for further domain reduction, thus ensuring the precise direction of the next search step. Experiments show that the proposed method outperformed other methods used in static test data generation. Specifically, it produces excellent results when variables are strongly related even when they are in equalities, and generation time increases stably and linearly with the increment of number of expressions including both equalities and inequalities.
목차
1. Introduction
2. Background
3. The Search Algorithm
3.1 State space search
3.2 Details of the proposed algorithm
4. Optimized Interval Arithmetic
4.1 The design of the algorithm
4.2 Case study
5. Experimental Analyses and Empirical Evaluations
5.1 Testing a composed test bed
5.2 Testing the performance of optimized interval arithmetic
5.3 Comparison with other static method
6. Conclusion
References