Name Disambiguation for Chinese Scientific Authors with Multi-Level Clustering

Name ambiguity arises when one retrieves publications written by distinct author entities who share the same name. During the past decades, a myriad of supervised and unsupervised methods have been proposed to resolve the ambiguity of author names and boost retrieval performance for digital libraries and other similar websites such as Citeseer1 and WanFang2. However, most of them either need large numbers of annotated data or cannot be scaled to suit massive data sets. In this paper, we propose an entirely unsupervised framework to achieve well-performed disambiguation, specifically, a multilevel clustering algorithm that builds a discipline tree in which paper and author entities are matched. To speed up the process of constructing discipline tree, we implement an efficient seeding algorithm for sequential k-Means and design a strategy for fast estimating k, the number of clusters. Experimental results show our framework is efficient for large scale data sets and works well for name disambiguation with respect to Chinese scientific authors.