问卷设计案例推理系统研究.DOC

问卷智能设计CBR系统及其关键技术研究 梁昌勇，顾东晓，黄燕 (合肥工业大学管理学院,合肥 230009) E-mail:dongxiaogu@yeah.net 摘要：针对问卷调查中问卷设计较大程度上依赖先前设计经验和知识的特点，将范例推理技术应用于问卷设计知识重用中的各个过程，提出了基于案例推理AHP、Delphi等常用权重调整方法操作不便、主观性大等问题，保证范例的检索质量。在此基础上建立了问卷智能设计系统，组织了原始范例库后进行了实验。结果表明， CBR问卷智能设计方法可较大程度上缩短设计时间、提高问卷设计效率和质量，有效地支持知识重用和创新。 Research of CBR System of Intelligent Questionnaire Design and the Key Technologies LANGI Changyong，GU Dongxiao，HUANG Yan (School of Management , Hefei University of Technology, Hefei 230009) Abstract: The design of questionnaire mostly depends on prior knowledge and experience. T. This paper puts forward an application structure of knowledge reusing method in designing questionnaire based on case-based reasoning. Then case representation, case retrieval by similarity measure and GATS, case optimization and case base reorganization are thoroughly explored. A CBR prototype system of questionnaire design is developed and some experiments are completed by using the platform. The results show its feasibility, validity, efficiency and quality. And the mechanism is based on the model of GA fitness function, using choosing, TS crossover and mutation to automatically optimize weight, which overcomes the disadvantages of common weight adjustment methods such as AHP and Delphi. A CBR prototype system of questionnaire design is developed and some experiments are finished by using the platform. The experiment result shows high efficiency, good quality of the technical platform, and effective support to the reuse of designing questionnaire knowledge and the innovation of knowledge management. Key words: cased-based reasoning; questionnaire design; knowledge reuse; hybrid retrieval process 1 引言 问卷调查是市调中最有效也是被经常使用的一种方法，在问卷调查中，问卷设计是非常重要的一个环节，决定着市调的成功与根据调查行业和调查方向的不同，问卷的设计在形式和内容上也有所不同，但是无论对于哪种类型的问卷来说，在设计过程中都必须要注意以下几个要点1）明确调查目内容设计时考虑数据统计和分析问题数量合理化、逻辑化、规范化GA的混合检索方法的有效性和效率进行了对比试验。 2 CBR机理及其在问卷设计中的应用框架 CBR是一种基于经验知识进行推理的人工智能方法。它根据目标范例的要求或提示，检索匹配源范例，并在其指导下形成求解方案[3，4]。CBR方法模仿人类决策过程中类比的认知方式