考虑重要性赋权的分部多关系聚类方法

Partition Multi-relational Clustering Based on Importance Weight of Relation

随着大数据时代的来临,实体之间的关系变得多种多样.实体和实体间的不同类型关系组成多关系网络,针对多关系网络中的实体进行聚类一直是热门的研究课题.本文综合现有多关系聚类方法的优势,提出了新的分部多关系聚类方法.首先通过对每个关系下的实体进行聚类,基于聚类结果对实体间关系进行重要性赋权,然后综合不同关系下实体关系的重要性权值,得到单关系网络,对该单关系网络进行聚类得到最终的聚类结果.最后对包括本文方法在内的不同聚类方法在多个公开数据集上进行了对比试验,验证了本文方法的有效性.本文方法对多关系聚类的准确度进行了提升,具有理论意义和应用价值.

With the advent of big data era, the relations between the entities become diverse. Different types of entities and relationships between entities form the multi-relational networks,clustering in the multi-relational networks has become a hot research topic. The paper takes advantages of the existing methods, and proposes a method of partition multi-relational clustering. At first, the method clusters the entities in every kind of relational network, and values the importance weight of relations according to the above result, and gets a single-relational network by considering the importance weight of relations in every kind of relational network. Clustering the entities in the single-relational network can get the final result. Finally,a comparative experiment is carried out on different clustering methods, including the method of this paper, which is based on a number of open data sets, and the validity of the method is verified. The method of this paper has an improvement on the accuracy of multi-relational clustering, which has theoretical significance and applicational value.