基于知识地图的知识推送方法研究  

Research on Knowledge Push Method with Knowledge Map

作　　者：渠国庆 熊峰[1] 牛倩[1] 吴祖伟[1] 吕北轩 

机构地区：[1]上海大学机电工程与自动化学院,上海市智能制造及机器人重点实验室,上海200072

出　　处：《计算机技术与发展》2017年第9期82-84,91,共4页Computer Technology and Development

基　　金：上海市科技计划项目(12DZ1505600)

摘　　要：对于知识密集型的高新企业来说,有效利用丰富且复杂知识的方法与途径,一直是企业发展的一个重要环节。随着推荐系统的出现与发展,催生了诸多经典推荐算法,并在电子商务和社交网络中得到了成功应用,但现有经典推荐算法并不适用于更加专业化的企业知识推送。为了提高企业知识与企业员工的匹配程度,使企业员工能更高效地利用企业知识,提出了一种改进的协同过滤推荐算法。该算法基于知识地图的知识推送方法,由知识地图的关联度得到更具体的关联关系,并综合考虑路径关联权值对推荐算法的影响,改进了相似度构成与计算方法,以相似度衡量用户之间的相似性,加强了用户之间的相似度的评估。理论分析表明,所提出的算法提高了知识推送的匹配程度和可行性,为企业知识推送提供了新的思路和途径。For knowledge-intensive and high-tech enterprises, the effective use of method with rich and complex knowledge has become an important part in development of enterprises. With the emergence and development of recommended system, a lot of classical recom- mendation algorithms have been produced and applied n e-commerce and social networks successfully, but all of the existed algorithms are not practically suited to more professional knowledge push for business enterprises. In order to increase matching degree between busi- ness knowledge and business employees and be efficient use of enterprise knowledge for employees, an improved collaborative filtering al- gorithm has been proposed. Based on an knowledge push method of knowledge map ,it obtains more specific relationship according to the relativity of knowledge map and improves the similarity construction and computation considering the impact of path associated weights on the recommendation algorithm. The similarity is utilized to measure the similarity between users, enhancement of similarity evaluation between users. Theoretical analysis shows that the matching degree and the feasibility of knowledge push has been increased by modified algorithm, which has provided a new way for future business knowledge push system in enterprises.

关 键 词：企业知识管理 知识地图 知识推送 协同过滤 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]