earticle

영어

With the rapid development of digital libraries, name disambiguation becomes more and more important technique to distinguish authors with same names from physical persons. Many algorithms have been developed to accomplish the task. However, they are usually based on some restricted preconditions and rarely concern how to be incorporated into a practical application. In this paper, name disambiguation is regarded as the technique of learning module integrated with a knowledge base. A network is defined for the modeling of publication information, which facilitates the representation of differenttypes of relations among the attributes. The knowledge base component serves as the user interface for domain knowledge input. Furthermore, this paper exploits a random walk with restart algorithm and affinity propagation clustering algorithm to finally output name disambiguation results.