A survey of author name disambiguation techniques: 2010–2016

Abstract

Digital libraries content and quality of services are badly affected by the author name ambiguity problem in the citations and it is considered as one of the hardest problems faced by the digital library researchers. Several techniques have been proposed in the literature for the author name ambiguity problem. In this paper, we reviewed some recently presented author name disambiguation techniques and give some challenges and future research directions. We analyze the recent advancements in this field and classify these techniques into supervised, unsupervised, semi-supervised, graph-based and heuristic-based techniques according to their problem formulation that is mainly used for the author name disambiguation. A few surveys have been conducted to review different techniques for the author name disambiguation. These surveys highlighted only the methodology adopted for author name disambiguation but did not critically review their shortcomings. This survey provides a detailed review of author name disambiguation techniques available in the literature, makes a comparison of these techniques at an abstract level and discusses their limitations.

The Impact of Name-Matching and Blocking on Author Disambiguation

“…They introduce a taxonomy that distinguishes among others between author grouping (resolving synonymy) and author assignment (homonymy). In a more recent comparison, Hussain and Ashger [8] discuss AND methods since 2010 and try to pinpoint strengths and weaknesses of the selected approaches. They use a different taxonomy focusing on the technology used (i.e., machine learning, graph-based, etc.…”

Incremental author name disambiguation using author profile models and self-citations

“…Current AND methods can be categorized into two groups: the first are the methods that resolve whole citations in BDs on new insertions called batch AND and the second are those methods which disambiguate only newly inserted citations called incremental AND. Similarly, Hussain et al in [6] proposed a taxonomy for existing AND methods and divided all methods into supervised [3,[11][12][13], unsupervised [14][15][16][17][18][19], semisupervised [20][21][22], graph-oriented approaches using graph models or social networks [1,[23][24][25], and string processing or heuristicbased methods [26][27][28]. In this section, we only overview the incremental AND methods.…”

“…In this section, we only overview the incremental AND methods. For detailed discussions about AND methods and AND data sets, interested readers are referred to our recent survey of these techniques presented in [6] and [29], respectively.…”

“…Although, several AND methods have been proposed (see, for instance, [6]), but the vast majority of those methods work for batch AND. To the best of our knowledge, there are only three incremental AND methods [7][8][9].…”

Strong Baselines for Author Name Disambiguation with and Without Neural Networks

“…But unfortunately, AND is not an elementary problem because distinct authors may share the same name, which is quite common for Asians, especially Chinese researchers [9], since different Chinese names will be the same when mapped to English (e.g., and share the same English name Wei Wang). The problem of disambiguating who is who dates back at least few decades, and it is typically viewed as a clustering problem and solved by various clustering models, such models have to answer two questions inevitably, that is how to quantify the similarity and how to determine cluster size [8]. Many existing literatures mainly focus on answering the first question, such as feature-based methods [12,13] and graph-based methods [3,16,20].…”