基于改进关联规则挖掘和距离聚类算法的知识融合  

Knowledge fusion based on improved association rule mining and distance clustering algorithm

作　　者：佟国香[1] 胡琪 杨培威 蒋伟 岳继光[2] TONG Guoxiang;HU Qi;YANG Peiwei;JIANG Wei;YUE Jiguang(School of Optical Electrical and Computer Engineering,University of Shanghai for Science and Technology,Shanghai 200093,China;School of Electronic and Information Engineering,Tongji University,Shanghai 200092,China)

机构地区：[1]上海理工大学光电信息与计算机工程学院,上海200093 [2]同济大学电子与信息工程学院,上海200092

出　　处：《智能计算机与应用》2023年第12期68-74,共7页Intelligent Computer and Applications

基　　金：国家重点研发计划项目(2018YFB1700902)。

摘　　要：针对产品全生命周期管理(PLM)中来自于不同阶段和不同领域的设计文档,本文提出一种基于实体抽取的改进关联规则挖掘与距离聚类相结合的知识获取与融合的算法。以汽车行业领域知识获取与融合为例,从相关文档中抽取出8 183组数据,220 941个实体,将各组数据的27个特征两两交叉并与原来的特征规则共同作为候选集,进行关联规则挖掘,并构造初步的领域知识库。通过定义知识库中各实体之间的距离,结合聚类算法减少冗余知识,再根据融合后的知识构建知识库。实验表明,该算法在一定程度上减少了知识模型的规模,提高了领域知识库中知识的价值密度。Aiming at the design documents from different generation stages and different fields in product lifecycle management(PLM),this paper proposed a knowledge acquisition and fusion algorithm based on improved association rule mining and distance clustering.Took the domain knowledge acquisition and fusion of automobile industry as an example.8183 groups of data and 220941 entities were extracted from relevant documents.The 27 features of each group of data were crossed in pairs and the original feature rules were combined as candidate sets to association rules mining and constructed a preliminary domain knowledge base.The distance between entities in the knowledge base was defined,combined with the clustering algorithm to reduce redundant knowledge,and the knowledge base was constructed according to the merged knowledge.The experimental results show that the algorithm reduces the scale of the knowledge model to a certain extent and improves the value density of knowledge in the domain knowledge base.

关 键 词：实体抽取 改进关联规则 距离聚类 产品全生命周期 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]