异质对象协同实体解析的联合聚类算法

实体解析问题是数据挖掘数据清理过程中的基本问题.异质网络数据的大量涌现,要求能够针对包含多种类型对象的数据同时进行实体解析.针对包含两种对象的实体解析问题,提出了一种基于联合聚类思想的协同实体解析算法.将两种对象分为决定对象和辅助对象,提出了一个基于联合聚类思想的两阶段协同实体解析框架,能够同时获得决定对象和辅助对象的各自聚类结果,其中每一个类包含的若干实体参考表示是对现实世界中同一实体的共同引用.最后对提出的算法进行了数值实验.

Co-clustering algorithm for collective entity resolution of multi-typed objects

Entity resolution is a fundamental issue in the process of data cleaning in data mining. With the advent of heterogeneous networks that consist of multi-typed objects, it is necessary to identify entities of different types simultaneously. In this paper, we propose a co-clustering algorithm for collective relational entity resolution in data set containing two types of entities. Specifically, we first classify objects into dominant objects and assistant objects according to their roles in entity resolution. Then, a two-phase framework for entity resolution of two object types based on co-clustering is presented, which obtains the object clustering results of each type simultaneously and considers references assigned to the same cluster as the same entity. A numerical example is also presented to illustrate the proposed new algorithm.