The Lean Model Of Software Maintenance And Data Mining Technology

Software crisis, formally has two aspects: quality crisis and cost crisis, while in nature is informational crisis in software developing process. It has two meanings: one is that information is in so large a scale that we can' t manage it well; the other is that logics in information are complicated and vary too often to understand. Model and algorithm, the two aspects of data mining technology can help us to solve those problems effectively: model helps us to find laws in information, while algorithm helps us to promote efficiency.For the sake of quality and cost in software maintaining with stipulations given, I introduced lean software maintaining model, which is a feed-back system made up of metrics and decisions making. In thisthesis I discussed several problems in the three factors of it and data mining technology using in them with details.For that stipulation metrics and implementing program metrics are separate in available software metrics technology, in this thesis I introduced R-Matrix and responsibility-based metrics, which is based on this idea: software is stipulations plus implementing program.As the second factor of the model, Comparison test is an automatic testing technique based on statistical models, which is generalized from validating-test technique, to solve this problem: how to choose an optimal solution from a couple of alternatives, with the potentials included.Lastly, I introduced the generalized refactor, the third factor of themodel. In this thesis I designed and implemented refactorring on R-Matrix using genetic algorithm.