基于多范例推理的小组软件过程规划算法  

Multicase-based reasoning algorithms for team software process planning

作　　者：刘群[1] 陈迎欣 

机构地区：[1]哈尔滨工程大学计算机科学与技术学院,黑龙江哈尔滨150001

出　　处：《哈尔滨工程大学学报》2007年第4期434-439,共6页Journal of Harbin Engineering University

摘　　要：由于软件开发小组和软件产品的不同,小组软件过程发生相应的改变,为了达到对软件过程由事后分析到事前规划的目的,必须要进行小组过程规划的研究.小组软件过程由活动组成,活动受角色、时间、数据、设备等方面的约束,软件过程的规划就是多约束满足问题.文中提出用问题归约策略与基于范例推理结合的方法解决这个问题.引出了范例粒度的概念,形成一种多范例推理的方法,对这种方法所涉及到问题和解决的办法做了详细的论述,并且验证该方法是有效的.由范例的重用体现出过程级上元素的重用,有利于软件过程的控制与优化.Every special project has a different development team and produces different software, so the team software process （TSP） should reflect those differences. Current studies on TSP shift post-analysis of software process to pre-planning, hence it is necessary to carry out the research work on TSP. TSP is constituted by activities, while activities are constrained by role, time, data, equipment, etc. TSP is designed to meet multi-constraint conditions. In this paper, a method is proposed that combines the strategy of problem reduction with case-based reasoning to give necessary flexibility to TSP. The concept of case granularity is introduced and a muhicase-based reasoning （MCBR） method is put forward. The paper discusses problems related to this method and gives the effective solutions to these problems. The validity of this method is verified as well. The reuse of the case embodies the reuse of the elements of process level, which is favourable for control and optimization of TSP.

关 键 词：软件过程 人工智能 规划算法 基于范例推理 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]