摘要
为了克服在基于案例推理中单一案例库检索的局限性,提出了基于贝叶斯信念网络的多案例库检索方法.该方法仅要求用户采用关键词对目标案例进行描述,通过贝叶斯信念网络计算关键词与各案例库的特征属性的匹配概率,以获得适应各案例库的完整的目标案例描述,然后对相应的案例库进行案例检索;将从多案例库中检索到的案例通过相似度线性归一处理后,结合目标案例描述的匹配概率获得与目标案例的最终相似度,统一排序后返回检索结果.最后通过航空企业的实际应用与性能分析进一步验证了多案例库检索方法的有效性.
To overcome the limitation of single case base retrieval,an approach to searching multiple case bases based on the Bayesian belief network is proposed.
出处
《北京工业大学学报》
EI
CAS
CSCD
北大核心
2012年第1期81-85,共5页
Journal of Beijing University of Technology
基金
国家自然科学基金资助项目(71101153)
教育部博士点基金资助项目(200800060005)
中国石油大学(北京)引进人才科研启动基金资助项目(QD-2010-06)
关键词
贝叶斯信念网络
基于案例推理
知识管理
are adapted to each case base to describe the target case.The matching degree between the Key words and the features are derived from the Bayesian belief network.The best match is used to search the case base.The liner normalization method is then used to normalize the similarity degree between the target case and the retrieved cases.The final similarity degree is derived using the normalized similarity degree and the matching degree of the target description.Consequently
retrieved cases ranked in final similarity degree are returned.Results of the application and experiment in the aeronautical enterprise show that the proposed approach is efficient to search multiple case bases. Key words： Bayesian belief networks
case-based reasoning
knowledge management
