A Trust-Based Approach to Estimating the Confidence of the Software System in Open Environments

Emerging with open environments, the software paradigms, such as open resource coalition and Internetware, present several novel characteristics including user-centric, non-central control, and continual evolution. The goal of obtaining high confidence on such systems is more difficult to achieve. The general developer-oriented metrics and testing-based methods which are adopted in the traditional measurement for high confidence software seem to be infeasible in the new situation. Firstly, the software development is changed from the developer-centric to user-centric, while user's opinions are usually subjective, and cannot be generalized in one objective metric. Secondly, there is non-central control to guarantee the testing on components which formed the software system, and continual evolution makes it impossible to test on the whole software system. Therefore, this paper proposes a trust-based approach that consists of three sequential sub-stages: 1) describing metrics for confidence estimation from users; 2) estimating the confidence of the components based on the quantitative information from the trusted recommenders; 3) estimating the confidence of the whole software system based on the component confidences and their interactions, as well as attempts to make a step toward a reasonable and effective method for confidence estimation of the software system in open environments. Supported by the National Basic Research 973 Program of China under Grant No.2009CB320702, the National Hi-Tech Research and Development 863 Program of China under Grant Nos. 2007AA01Z178 and 2007AA01Z140, the National Natural Science Foundation of China under Grant Nos. 60603034, 60721002 and 60736015, and the Jiang Su Natural Science Foundation under Grant No. BK2008017.