Approach for Name Ambiguity Problem Using a Multiple-Layer Clustering

Name ambiguity refers to the problem of attributing a publication to a proper author. This is a common issue in digital library. It is a difficult problem as the same author´s name may be written in different ways and different authors may share the same name. In this paper, we examine a multiple-layer clustering approach which is based on a limited amount of associated information with each publication. It combines the Package-Merge algorithm, pattern-matching extraction methods, as well as a fuzzy logic rule based concept. This experimental study uses the DBLP collection as a case study, and the three attributes used are email addresses, the co-authorship relationship and paper title similarity. Our experiments show that this approach can distinguish authors and classify papers on the test dataset more accurately than the previous studies.