中文摘要：

针对自然语言描述的软件缺陷报告常会出现不完整、二义性、抽象层次混杂的问题，面向黑盒测试发现的缺陷，提出了一种图形化的软件缺陷描述语言——软件缺陷描述图（SDDG）．首先采用扩展巴科斯范式（EBNF）对SDDG进行形式化描述，然后以因果图为基础，定义了描述中涉及的各种图形符号．通过理论分析论证、描述实例对比以及实验验证表明：与传统的自然语言描述以及基于XML的描述相比，SDDG清晰明了、重点突出，提高了软件缺陷报告的可读性、完整性以及可重现性，减少了测试人员和开发人员的理解不一致现象；该方法不仅提高了双方的沟通效率，同时为软件缺陷分类和分析提供了更好的原始数据．

英文摘要：

Ambiguity, incompleteness and confusion abstraction layers in software black-box defect report described in natural language were focused. Aimed at the detected defects in black-box testing, a graphed method named software defect description graph （SDDG） was proposed. The SDDG wasformally defined with extended Backus-Naur form （EBNF）, and the graphical symbols were defined based on cause-effect graph. Theoretical analysis, case study and experimental verification show that SDDG is simpler and clearer than defect description in natural language and XML, and it can improvethe readability, integrity and reproducibility of defect report, and it can also reduce the inconsistent understanding caused by ambiguity of natural language. The method improves the communication efficiency of tester and programmer, meanwhile it provides better raw defect data for software defect classification and analysis.