Publication

Abstract : Software maintenance incurs a significant cost during the life cycle of any software product. This is due to multiple factors such as addition of features demanded by new requirements, modification of existing features triggered by fixing of defects in the code and extending product support to newer platforms. These factors cause the testing phase to be repeated each time thereby increasing the maintenance cost. The quality of the code compounds to the maintenance cost since poorly written code are hard to comprehend and work with. This paper proposes a methodology to reduce the testing effort through successive releases by determining a set of most significant functionalities by computing scores based on the usage frequency, code maintainability and change vulnerability. The scores are recomputed as the software evolves over successive releases to keep the testing effort minimal.