QQ客服 官方微博 反馈留言

题名	基于过程资产库挖掘的过程Agent自动创建方法
作者	黄辉
学位类别	硕士
答辩日期	2011-05-30
授予单位	中国科学院研究生院
授予地点	北京
导师	王青
关键词	计算机软件::软件工程
其他题名	A Process Asset Library Mining Based Method for Creating Process-Agents
学位专业	计算机软件与理论
中文摘要	软件可信性是当今软件工程领域面临的一个挑战性难题，而作为保证软件可信的重要方法，可信软件过程也成为当前最为活跃的研究内容之一。软件过程的可信性是一个软件过程能够生产出满足用户需求的软件产品的信心程度。但是，软件过程是动态的，并且是高度依赖于人的，因而，一个软件过程的性能不仅仅取决于过程本身定义的好坏，而且还取决于过程的执行者，即软件组织的人力资源，他们的能力将最终会影响到软件的可信性。与传统制造工业中的机器不同，人的能力是动态变化，对人的能力尤其是执行过程的能力的评价特别困难。因此，提供一种能够自动提取并评价人力资源能力的方法特别重要。鉴于软件过程对人的高度依赖性，中国科学院软件研究所提出了一种基于组织实体能力的软件过程建模（OEC-SPM）方法。该方法充分考虑到人力资源能力的因素，并将具有相似能力的一组人封装为一个过程Agent。一个过程Agent的构成包括行为引擎和知识结构两部分，其中引擎提供了过程Agent的行为机制，知识结构则包含了过程Agent所具有的知识，包括描述性知识、过程性知识和经验库。知识结构能够全面客观的描述过程Agent的能力，并且通过创建的过程Agent，也能够实现软件组织人力资源的组织和管理。基于过程Agent的知识结构，经过引擎的推理和协商，可以为组织自动生成可靠的软件开发过程，从而提高软件开发的效率以及过程的可信性。过程Agent是组织人力资源、评价人力资源能力，并且将其能力和软件过程连接起来的一种合理的方法。本文在分析了软件组织过程资产库的基础上，提出了一种基于过程资产库挖掘的过程Agent创建方法。该方法在历史项目数据的基础上，通过数据预处理和数据挖掘来创建过程Agent，从而实现对软件组织人力资源的管理和重新组织，为制定项目计划、组建项目团队、调度人力资源等提供支撑。并且考虑到软件过程的动态性以及人力资源的可变性，该方法提供了三种创建过程Agent的场景，可以根据人力资源能力的变化增量的创建过程Agent。然而在分析过程资产库时，本文还发现生产率数据缺失十分严重，生产率是过程Agent经验数据的重要组成部分，其缺失将会严重影响过程Agent的质量，因而本文提出了一种能够自动修复缺失和不准确生产率的方法，修复后的生产率数据将作为过程Agent知识的一部分，用于评价其能力。最后，本文还将提出的方法在一个软件开发组织进行了应用。基于该组织真实的开发数据，实验表明本文的方法能够修复缺失的生产率数据，并且在处理后的数据的基础上，能够自动高效的提取高质量的过程Agent。
英文摘要	Software trustworthiness is a challengeable problem in software engineering field. Trustworthy software process which is an important approach to access software trustworthiness is one of most active research topics in recent years. Software process trustworthiness is the degree of confidence that a software process produces expected trustworthy work products that satisfy their requirements. But software processes are dynamic and highly people-dependent. Therefore, the performance of software processes not only relies on processes themselves, but also depends on the developers’ capabilities, which will finally impact the software trustworthiness. Furthermore, capabilities of the developers also change with time, which is very different from machines in traditional manufacturing. Evaluation of human resources’ capabilities, especially their process capabilities, is extremely hard. Thus it is important to provide a method to extract and evaluate human resources’ capabilities automatically.In view that software process is highly people-dependent, Institute of Software Chinese Academy of Sciences (ISCAS) provided an Organization-Entity Capability based Software Process Modeling (OEC-SPM) method, in which capabilities of human resources are fully considered and a group of human resources with certain capabilities is encapsulated as a Process-Agent. A Process-Agent consists of two parts: Engine and Infrastructure. The Engine provides an acting mechanism, while the Infrastructure contains the knowledge a Process-Agent has, including descriptive knowledge, process knowledge and Experience Library. Therefore, the Infrastructure can describe human resources’ capabilities comprehensively and objectively, and human resources also can be well organized and managed through the creation of Process-Agent. Furthermore, based on the reasoning and cooperation in Engine, software project process can be established automatically, which will improve the efficiency of software development and the trustworthiness of software process. Thus Process-Agent is a good way to organize human resources, evaluate their capabilities and connect them to software process.Based on the analysis of software organization process asset library, this thesis proposes a process asset library mining based method for Process-Agent creation. On the basis of the history project data in process asset library, the method creates Process-Agents through data mining to implement human resource management and reorganization, which can provide support for project plan making, project team construction, and human resource scheduling and so on. Furthermore, considered the dynamic of software process and the variability of human resource, the method provides three scenarios in which Process-Agents can be created incrementally according to the variation of human resources’ capabilities. Besides, during the analysis of process asset library, it is found that the productivity which is a crucial metric for human resource evaluation is imprecise, thus this thesis also proposes a method for precise productivity estimation. The refined productivity data will be a part of Process-Agents.At last, this thesis applies the two methods in a software development organization. Based on the real development data, the experiment shows that the imprecise productivity data can be refined and the Process-Agents can be created with high quality.
语种	中文
公开日期	2011-06-08
内容类型	学位论文
源URL	[http://124.16.136.157/handle/311060/10215]
专题	软件研究所_互联网软件技术实验室 _学位论文
推荐引用方式 GB/T 7714	黄辉. 基于过程资产库挖掘的过程Agent自动创建方法[D]. 北京. 中国科学院研究生院. 2011.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。