中文摘要：

传统的软件聚类方法没有考虑软件实体间存在一些无法通过静态分析手段获取的关系（即演化依赖关系）,这将导致聚类后的软件可能不符合＂高内聚、低耦合＂特征.为了解决上述问题,提出了将软件演化信息纳入软件聚类过程的策略,并在扩展的Java依赖模型的基础上,实现了一个基于模拟退火的软件聚类算法.实验结果表明：该方法能达到提高软件聚类准确度的目的.

英文摘要：

Software clustering can be used to solve the software partitioning problem and realize the software modularization. However,traditional software clustering methods have not considered the potential relation between software elements,which cannot be identified by using the static analysis method. So it may lead to software not meet the ＂ high cohesion,low coupling＂ feature after clustering. In order to solve the above problem,a strategy by introducing the software evolution information into the software clustering process,and propose a software clustering algorithm based on the extended Java dependence model and simulated annealing idea have been proposed. Experiments show that this method can improve the accuracy of software clustering.