Learning CNF Blocking for Large-scale Author Name Disambiguation

Abstract

Author name disambiguation (AND) algorithms identify a unique author entity record from all similar or same publication records in scholarly or similar databases. Typically, a clustering method is used that requires calculation of similarities between each possible record pair. However, the total number of pairs grows quadratically with the size of the author database making such clustering difficult for millions of records. One remedy is a blocking function that reduces the number of pairwise similarity calculations. Here, we introduce a new way of learning blocking schemes by using a conjunctive normal form (CNF) in contrast to the disjunctive normal form (DNF). We demonstrate on PubMed author records that CNF blocking reduces more pairs while preserving high pairs completeness compared to the previous methods that use a DNF and that the computation time is significantly reduced. In addition, we also show how to ensure that the method produces disjoint blocks so that much of the AND algorithm can be efficiently paralleled. Our CNF blocking method is tested on the entire PubMed database of 80 million author mentions and efficiently removes 82.17% of all author record pairs in 10 minutes.

Whois? Deep Author Name Disambiguation using Bibliographic Data

“…Most of the studies treat the problem of author name ambiguity as an unsupervised task [18,36,17,17,26] using algorithms like DBSCAN [17] and agglomerative clustering [31]. Liu et al [21] and Kim et al [18] rely on the similarity between a pair of records with the same name to disambiguate author names on the PubMed dataset. Zhang et al [36] used Recurrent Neural Network to estimate the number of unique authors in the Aminer dataset.…”

“…These approaches rely on the matching between publications and authors which are verified either manually or automatically. Unsupervised approaches [21,18,5] have also been used to assess the similarity between a pair of papers. Other unsupervised approaches are also used to estimate the number of co-authors sharing the same name [36] and decide whether new records can be assigned to an existing author or a new one [26].…”

Whois? Deep Author Name Disambiguation Using Bibliographic Data

“…Most of the studies treat the problem of author name ambiguity as an unsupervised task [6,9,9,13,15] using algorithms like DBSCAN [9] and agglomerative clustering [21]. Liu et al [12] and Kim et al [13] rely on the similarity between a pair of records with the same name to disambiguate author names on the PubMed dataset. Zhang et al [15] used Recurrent Neural Network (RNN) to estimate the number of unique authors in the Aminer dataset.…”

“…In this work, we collected our dataset from the DBLP bibliographic repository 13 . The DBLP version of July 2020 contains 5.4 million bibliographic records such as conference papers, articles, thesis, etc., from various fields of research.…”

“…These approaches rely on the matching between publications and authors which are verified either manually or automatically. Unsupervised approaches [12][13][14] have also been used to assess the similarity between a pair of papers. Other unsupervised approaches are also used to estimate the number of co-authors sharing the same name [15] and decide whether new records can be assigned to an existing author or a new one [6].…”