The increasing demand for complex applications in diversified areas imposes a great challenge on developing software systems to deal with imprecise and uncertain information. One of the foci of the recent developments in software engineering is the investigation of computational intelligence ~CI! for software engineering to address the ever-increasing complexity and size of software systems, and the imperfect information inherited in nature. These treatments enable the extension of CI to various phases in software life cycle along three dimensions:
β’ managing fuzziness resided in the requirements, including the formulation of imprecise requirements and the trade-off analysis for conflicting requirements β’ coping with fuzzy objects and imprecise knowledge, including the modeling of objects and their relationships
β’ handling uncertainty encountered in quality prediction, including the utilization of learn- ing mechanisms to develop general prediction models for spotting the most troubleprone modules early in the software life cycle In this special issue, we are featuring three articles devoted to the extension of CI to software engineering.
Marek Reformat presents a comprehensive meta-model for the prediction of the number of defects. It dwells on evidence theory and a number of fuzzy-based models developed using different techniques applied to different subsets of data. In this article, attention has been directed to the prediction of the number of defects in a given software system. Fuzzy models, extracted knowledge, and results of the proposed prediction system have been described and investigated. The models are developed using an evolutionary-based approach with different objectives applied to different subsets of data. Evidence theory and belief function values assigned to