학술논문
基于IG-NRS与ICK双向压缩的KMS自学习案例知识匹配 / KMS self-learning case knowledge matching based on IG-NRS and ICK bidirectional compression
Document Type
Academic Journal
Author
Source
计算机应用研究 / Application Research of Computers. 41(2):393-400
Subject
Language
Chinese
ISSN
1001-3695
Abstract
案例知识匹配可有效缓解知识过载问题,确保知识应用水平.针对知识管理系统自学习案例知识匹配的冗余性问题,提出了一种基于IG-NRS与ICK的双向压缩方法.该方法首先设计NRS的改进模型IG-NRS,据此约简案例知识属性集,实现邻域决策系统的纵向压缩;在此基础上,通过引入谱聚类判别并剔除不一致案例知识实现其横向压缩;再藉由知识视图相似度锁定目标案例知识簇与最相似案例知识,从而确定知识匹配结果.在多个UCI数据集上的实验结果表明,该方法能有效减少知识管理系统自学习案例知识的冗余,取得更高的知识匹配效率和有效性.
Case knowledge matching can effectively alleviate the knowledge overload problem and ensure the level of know-ledge application.Aiming at the redundancy problem of self-learning case knowledge matching in knowledge management sys-tems,this paper proposed a bidirectional compression method based on IG-NRS and ICK.The method firstly designed an im-proved model of NRS,called IG-NRS.Accordingly,it approximated the set of case knowledge attributes to achieve the vertical compression of the neighborhood decision system.On this basis,it realized horizontal compression by introducing spectral clus-tering discrimination and eliminating the inconsistent case knowledge.And then it determined the knowledge matching results by locking the target case knowledge clusters and the most similar case knowledge through the similarity of knowledge views.Experimental results on several UCI datasets show that this method can effectively reduce the redundancy of self-learning case knowledge in knowledge management systems and achieve higher knowledge matching efficiency and effectiveness.
Case knowledge matching can effectively alleviate the knowledge overload problem and ensure the level of know-ledge application.Aiming at the redundancy problem of self-learning case knowledge matching in knowledge management sys-tems,this paper proposed a bidirectional compression method based on IG-NRS and ICK.The method firstly designed an im-proved model of NRS,called IG-NRS.Accordingly,it approximated the set of case knowledge attributes to achieve the vertical compression of the neighborhood decision system.On this basis,it realized horizontal compression by introducing spectral clus-tering discrimination and eliminating the inconsistent case knowledge.And then it determined the knowledge matching results by locking the target case knowledge clusters and the most similar case knowledge through the similarity of knowledge views.Experimental results on several UCI datasets show that this method can effectively reduce the redundancy of self-learning case knowledge in knowledge management systems and achieve higher knowledge matching efficiency and effectiveness.