Adaptive Random Testing (ART) aims at improving detection effectiveness by evenly distributing test cases over the whole input domain. Many ART algorithms introducing clustering techniques (such as k-means Clustering) have been proposed to achieve an even spread of test cases. Though previous studies have demonstrated that ART with k-means clustering could achieve a good enhancement in testing effectiveness, k-means clustering is limited by the value of k, which will have a great impact on the test effectiveness. To improve the testing effectiveness of these techniques for object-oriented software, in this paper, we propose an approach named Determination Method of Optimal k-value based on the Experimental Process (DMOVk-EP) to determine the optimal k-value of k-means clustering and make the ART algorithms using k-means clustering technique achieve the best fault detection capability. The proposed method consists of two parts, one is a solution model for k based on the experimental process, and the other is an optimal k-value algorithm based on the presented model. We integrate this method with k-means clustering in ART and apply it to a set of open-source programs, with the experimental results showing that our approach obtains much more appropriate k, and also achieves much better testing effectiveness than other related methods.

Software vulnerability analysis is one of the critical issues in the software industry, and vulnerability classification plays a major role in this analysis. A typical vulnerability classification model usually involves a stage of term selection, in which the relevant terms are identified via feature selection. It also involves a stage of term weighting, in which document weights for the selected terms are computed, and a stage for classifier learning. Generally, the term frequency-inverse document frequency (TF-IDF) is the most widely used term-weighting method. However, empirical evidence shows that the TF-IDF is plagued with issues pertaining to its effectiveness. This paper introduces a new approach for vulnerability classification, which is based on term frequency and inverse gravity moment (TF-IGM). The proposed method is validated by empirical experiments using three machine learning algorithms on ten publicly available vulnerability datasets. The result shows that TF-IGM outperforms the benchmark method across the applications studied.