【作者】 席运江；

【导师】 党延忠；

【作者基本信息】 大连理工大学 ， 管理科学与工程， 2007， 博士

【摘要】 在组织知识管理中，面临的首要问题是如何描述组织知识，并据此进行各种知识管理活动。这一问题包含如下内容：“组织中有哪些知识，这些知识是什么，有多少，存储在哪里?”。可以看出，这些问题都是组织知识管理中的基础性问题，只有把这些问题理清，才可以明了组织知识管理的对象和内容，并在此基础上开展各项知识管理工作。然而，由于知识的模糊性和复杂性，至今尚无较完备的可描述组织知识的模型，尤其是缺少可满足组织知识管理要求的、相对定量的表示模型，这也导致许多知识管理工作只能做到粗略的和简单的管理，难以深入细致的进行。针对这些问题，本文在对组织知识的构成进行分析的基础上，结合复杂网络研究的若干方法和技术，对组织知识的结构建模和存量建模等问题进行了研究，并针对组织知识管理中的若干问题，对模型的应用进行了探讨。本文的主要研究内容及成果如下：1、组织知识结构的网络表示模型及建模方法。在对组织知识构成进行分析的基础上，并结合组织知识的存储情况，提出了组织知识的K-K网络模型和KSN(知识超网络)模型。其中，K-K网络模型可以描述组织知识的结构和构成，也可以用于表示组织的领域知识、组织成员的个人知识等；KSN模型将组织知识及其存储载体集成在一起，在描述组织知识结构的同时，可同时反映组织知识的存储分布情况。利用K-K网络模型和KSN模型，可较全面的反映组织知识的结构、构成及分布情况，并可用图形等形式表示出来，因而具有直观、简便等特点。在论文中，还对K-K网络和KSN网络的建模方法进行了探讨。为解决大批量文档的K-K网络建模问题，文中提出了基于词形关联网络的文本知识网络建模方法，该方法可以通过计算机辅助建立K-K网络模型。2、组织知识存量的加权网络表示模型在K-K网络模型及KSN模型基础上，提出了组织知识存量表示的WKN(加权知识网络)模型及WKSN(加权知识超网络)模型。其中，WKN模型既可以表示个人知识存量，也可以表示组织知识存量。在个人WKN模型中，节点权重反映了个人对知识点的掌握程度，用以表示个人知识存量；在组织WKN模型中，知识节点的权重分为两类：生命载体存量权重值和物质载体存量权重值，分别用以表示组织知识在人脑中和各类物质载体中的存量；在WKSN模型中，除了知识节点具有权重以表示组织知识存量外，在知识节点与各存储载体节点之间的边也具有权重，用以表示组织知识点在各存储载体中的存量。利用WKN和WKSN模型，可表示组织知识的存量构成及分布情况，为组织知识的定量分析和管理奠定基础。3、基于WKN模型的个人及群体知识结构分析方法在WKN模型基础上，对个人及群体知识结构的分析方法进行了研究。主要内容包括：个人及群体领域知识表示方法；个人及群体领域知识的广度、深度分析方法；交叉领域知识分析方法。上述方法，可初步实现个人及群体知识结构的定量化表示和分析。4、基于WKSN模型的组织人才流失问题分析及评价方法在WKSN模型基础上，对由于人才流失而导致的组织知识安全问题进行了研究。基本思路是，首先将组织成员知识表示为一个WKSN模型，然后通过分析WKSN网络的鲁棒性来分析由于人才流失导致的组织知识安全问题。在进行WKSN网络鲁棒性分析时，综合考虑了知识和知识主体两种类型的节点及其之间的相互关系。在节点移除时，提出了一种关联节点移除的方法来研究WKSN网络的鲁棒性，并在此基础上，提出了评价人才流失问题的若干指标，如专有知识率、专有知识加权比率、知识网络抗毁性、核心领域知识网络抗毁性等指标。根据这些指标可以对组织的人才流失问题进行分析和评价，例如可应用于组织知识资源的抗毁性评估、发现易流失知识以及评价组织成员的知识重要性等方面。5、基于WKSN模型的组织知识定位方法在WKSN模型基础上，对组织知识的定位方法进行了研究。在方法中，首先将待查找知识点的存量及结构表示为组织WKN网络的一个子网，然后在WKSN网络中搜索与待检索知识点相关的人及物质载体，同时将搜索结果表示为组织WKSN的一个子网，最后根据子网中的知识点存量分布情况对检索结果进行排序，使与待检索知识点相关程度最高的人或物质载体排在前面，从而实现快速准确的知识定位，为组织知识使用提供便利。需要说明的是，在本论文中，主要集中于对组织知识的表示及分析方法进行研究，但对具体的知识获取方法(例如隐性知识获取、文本知识获取)不作深入探讨。此外，在论文中，实例的作用主要是用来说明模型及方法的过程和特点，因此，在实例中采用了很多简化的处理方法，例如未在实例中考虑隐性知识及其他非文本知识等。上述简化处理方法，仅可满足研究的需要。如果要在应用中发挥作用，还必需和其他的研究相结合，例如知识获取及计量方法(尤其是隐性知识的获取和计量方法)、文本挖掘方法、ontology、领域知识词典等，但限于篇幅在本论文中不进行深入讨论。更多还原

【Abstract】 How to represent organizational knowledge is a central and principal problem in organizational knowledge management. The problem includes such following questions: what is organizational knowledge? How about its composition? How much knowledge in an organization? And where is it? Obviously, the problem is very principal and important in organizational knowledge management, which can help us know the contents or objects of organizational knowledge management. However, because of the ambiguity and complexity of knowledge, it’s really very difficult to represent organizational knowledge, especially to construct a quantitative model to represent and analyze organizational knowledge intensively and validly.To resolve these problems, based on the analysis of its structure and compositions, the dissertation aims to propose some methods to modeling the structure and knowledge stocks of organizational knowledge by using network models. The dissertation also suggests some possible applications of the models proposed in the paper. The results and achievements can be summarized as follows:1. The network models and their modeling methods to the structure of organizational knowledge.Based on the analysis of the structure and compositions together with the storage media of organizational knowledge, the knowledge network model (K-K network) and knowledge supemetwork model (KSN) are proposed in the paper. The K-K network model can represent the structure and compositions of organizationaI-knowledge and some of its knowledge fields, and can also represent the structure & compositions of individual knowledge. The KSN model, in which organizational knowledge and its storage media are integrated into a supernetwork, can not only represent the structure of organizational knowledge, but also can represent how it is stored in different storage media. The KSN model can also be used to describe the activities of organizational knowledge management. The modeling methods of K-K network and KSN are also discussed in the dissertation. To deal with large quantity of documents, a modeling method to document knowledge in organization is proposed based on Word-Relation Network analysis, which can be used to construct K-K network with assistant of computers.2. The weighted network models to organizational knowledge stocks. Based on the K-K network and KSN model, the weighted knowledge network model (WKN) and weighted knowledge supernetwork model are proposed to represent the organizational knowledge stocks. The WKN can represent individual and organizational knowledge stocks. In the WKN model for individual, the weights of nodes donate to what degree the knowledge points are mastered by an individual, while in the WKN model for organization, the weights of nodes have 2 types: one is used to donate the knowledge stocks in individual brains, the other is in material media. In WKSN model, besides the weights of knowledge nodes, the edges from individuals and material media to knowledge nodes also have weights, which donate the stocks that knowledge points are stored in individual brain or material media..3. The method to analyze the knowledge structure of an individual or a group based on WKN model.Based on the WKN model, the methods to represent the domain knowledge, to measure the extensiveness and the intensiveness of the domain knowledge of an individual or a group are presented, and some special features of the interdisciplinary knowledge are also discussed in the dissertation.4. The analyzing & measuring methods to the intelligence loss problem of an organization based on WKSN.The method is focused on the knowledge security in organization caused by the resignation of its members. Firstly the WKSN model is proposed in which all the members and their individual knowledge are integrated together according to the relation mappings between them. Based on the WKSN model, a combined node removal method is proposed. To measure the robustness of knowledge network, some signals such as the unique knowledge proportion, the weighted proportion of unique knowledge, the resilience of knowledge network, the resilience of core field knowledge network, are proposed and analyzed. The method proposed in the dissertation can successfully analyze and measure the robustness of knowledge network, and can also be applied to assess the security of knowledge resource in an organization, to discover the knowledge points that are easy to be lost, ancl to evaluate the importance of each member in the organizational knowledge.5. The method to locating organizational knowledge based on WKSNBased on the WKSN model, a method to searching & locating organizational knowledge is proposed. Firstly the knowledge points to be located can be represented as a subnet of WKN, then all individuals and material storage media that are related to the knowledge points can be retrieved and can be represented as a subnet of WKSN. According to the weighted from the individuals and material storage media to specific knowledge points, the result can be ordered so as to be convenient to locate the organizational knowledge. Noted that the dissertation aims at the methodology of representation and analysis of organizational knowledge instead of the knowledge acquisition methods such as tacit knowledge acquisition or text knowledge acquisition. Additionally, the instance in the dissertation is merely to clarify the processes and characters of the models or methods, therefore some processes such as the acquisition of tacit knowledge and text knowledge are simplified or neglected. The simplifid processes may be enough for research while not for practice. If the methods are going to be applied in practice, some supplementary methods should be taken into account together such as knowledge acquisition methods, knowledge measurement methods, text mining methods, ontology, fields dictionary etc.更多还原