中文摘要：

针对软件行为相关性提出一种基于HMM的分析方法，以软件行为属性：参数、上下文环境、动作、状态、接口和端口为可观测状态集，构建了行为相关性分析算法．首先，通过可观测序列训练出HMM数学模型，然后根据模型和可观测序列计算隐藏的软件系统行为状态序列，最后用软件部件状态向量表示软件行为状态，通过软件系统状态迁移而引起的部件状态向量的迁移分析出部件之间的相关性．在文章的最后通过仿真实验检验了该软件行为相关性分析方法的可行性和有效性．

英文摘要：

This paper presents a method to analyze the behavior of software relevance based on HMM and constructs a behavior of software relevance algorithm, taking the software behavior properties： parameters, context, actions, interfaces and ports as the observa- ble state sets. First,the HMM model is trained by a observable sequence;And then,according to the trained HMM model and the ob- servable sequences,the hidden software system behavior state sequences are calculated ;Finally, the state vectors of the software com- ponents are used to express the behavior＇s state of software system,at the same time,the relevance between components, which is caused by the migration of the components state vectors,is analyzed. At the end of the paper,the feasibility and effectiveness are tested by simulation experiments for the method.