One adaptive modeling with markov-chain decision tree in application of independent combinational components installing test

Abstract

The software functional test goal is to detect and remove as many defects as possible with a series of test case. In order to solve the problem of paradox of defects coverage and time cost in installation of independent combinational components(ICCI), one adaptive modeling algorithm with Markov-chain decision tree based on some necessary assumptions is proposed in this paper compared to the traditional matrix methodology. After comparing two searching strategies called two- binary search and bottom-totop respectively, we also do some edges cunning work according to improved understanding of correct probability of software product during testing and defect verification. Combining the merits of the two strategies with layer controlling, empirical results with one of our product Tivoli Integrated Portal is also collected to demonstrate the effectiveness of such solution. © 2009 IEEE.