基于双维压缩与综合活性的案例知识进化研究

Research on Case Knowledge Evolution Based on Two-Dimensional Compression and Integrated Activity

摘要:

知识作为企业价值创造的核心资源，其生命周期在加速缩短，导致知识低效、失效、冗余等问题日益突出。知识进化可有效缓解前述问题，可提升案例库质量、改善知识应用效用。基于双维压缩与综合活性的案例知识进化方法，首先采用C4.5-NRS算法约简案例属性集，完成案例库纵向压缩，减少后续计算的工作量和冗余属性的影响；其次，基于改进K-means聚类横向压缩案例空间，并结合聚类中心信息熵圈定待进化案例簇；而后，基于时效活性、应用活性、稀缺活性、熵活性等指标，得到待进化案例知识的综合活性；最后，依据既定阈值确定待进化案例的进化操作。算例结果表明，进化后案例库的平均活性和运行效率均有明显提高。其中，对进化空间实施双维空间压缩，提高了计算效率；通过在保留和删除的二元化操作基础上增加更新、休眠操作，增强活性中等的案例的活性，确保案例库的存量与质量的协同发展。

Abstract:

As the core resource of enterprise value creation,knowledge's life cycle is accelerating and shortening,which leads to the problems of knowledge inefficiency,invalidation and redundancy. Knowledge evolution can effectively alleviate the above problems,improve the quality of case base and improve the application utility of knowledge. In this paper,a case knowledge evolution method based on double-dimensional compression and comprehensive activity is proposed. Firstly,the C4.5-NRS algorithm is used to reduce the case attribute set,and the case base is vertically compressed to reduce the workload of subsequent calculation and the impact of redundant attributes. Secondly,the case space is horizontally compressed based on the improved K-means clustering,and the case clusters to be evolved are delineated by combining the information entropy of the cluster center. Then,the comprehensive activity of the case knowledge to be evolved is obtained based on the indicators of timeliness activity,application activity,scarcity activity and entropy activity. Finally,the evolutionary operation of the case to be evolved is determined according to the established threshold. The simulation results show that the average activity and operation efficiency of the evolved case base are significantly improved. Among them,the double-dimensional spatial compression of the evolution space improves the computing efficiency. By adding update and sleep operations on the basis of the binary operation of retention and deletion,the activity of the cases with medium activity is enhanced to ensure the coordinated development of the stock and quality of the case base.