Ranking authors in academic social networks: a survey
| Date | 19 March 2018 |
| DOI | https://doi.org/10.1108/LHT-05-2017-0090 |
| Pages | 97-128 |
| Published date | 19 March 2018 |
| Author | Tehmina Amjad,Ali Daud,Naif Radi Aljohani |
| Subject Matter | Library & information science,Librarianship/library management,Library technology,Information behaviour & retrieval,Information user studies,Metadata,Information & knowledge management,Information & communications technology,Internet |
Ranking authors in academic
social networks: a survey
Tehmina Amjad and Ali Daud
International Islamic University, Islamabad, Pakistan, and
Naif Radi Aljohani
Faculty of Computing and Information Technology,
The University of King Abdulaziz, Jeddah, Saudi Arabia
Abstract
Purpose –This study reviews the methods found in the literature for the ranking of authors, identifies the
pros and cons of these methods, discusses and compares these methods. The purpose of this paper is to study
is to find the challenges and future directions of ranking of academic objects, especially authors, for future
researchers.
Design/methodology/approach –This study reviews the methods found in the literature for the ranking
of authors, classifies them into subcategories by studying and analyzing their way of achieving the
objectives, discusses and compares them. The data sets used in the literature and the evaluation measures
applicable in the domain are also presented.
Findings –The survey identifies the challenges involved in the fieldof ra nkingo fauthors and future directions.
Originality/value –To the best of the knowledge, this is the first survey that studies the author ranking
problem in detail and classifies them according to their key functionalities,features and way of achieving the
objective according to the requirement of the problem.
Keywords Academic social networks, Author ranking, Expert finding, Learning-based ranking,
Link analysis, Text similarity ranking
Paper type Literature review
1. Introduction
With the emergence of social network, the world has become a very small place where
people are connected to each other via satellite channels, wireless communications 3G/4G
networks and many more. We can define a social network as a network within which
individuals and/or organizations are arranged as nodes (called actors) and are largely
interconnected via edges signifying various relationships, for example, co-authorship,
citations, references, recommendation, friendship, likes and dislikes, etc. Representative
social networks that are very popular include Twitter, Facebook, Flickr, Instagram,
YouTube, etc. Social networks are often represented as graph structures to facilitate mining
and analysis of the networks.
Academic social networks (ASNs) are a subclass of social networks with scientific
researchers as the main actors who collaborate in a research and appear as co-authors of
publications. Such networks are now materialized on the internet and are well supported by
various social networking service platforms. Many online publication repositories, such as
Citeseer[1] and DBLP[2] are good examples of materialized ASNs on the Internet. They are
frequently used for various mining tasks such as author ranking and expert
recommendation. With production of a large number of scientific articles, finding
relevant information has become a problem in recent years. With an exponential increase in
the size of the data, growing computational powers and economical storage mechanism,
the problem of finding relevant information has gathered attention of the researchers.
With an increase in the size of scholarly data, ranking in ASNs has also become an integral Library Hi Tech
Vol. 36 No. 1, 2018
pp. 97-128
© Emerald PublishingLimited
0737-8831
DOI 10.1108/LHT-05-2017-0090
Received 6 May 2017
Revised 12 September 2017
Accepted 9 November 2017
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/0737-8831.htm
The work is supported by Higher Education Commission (HEC), Pakistan startup research grant under
Interim Placement of Fresh PhDs program 2011, and the Indigenous Ph.D. Fellowship Program.
97
Academic
social
networks
part of these networks. These methods are required for expert finding, research grant
recommendations, finding relevant reviewers and members for editorial panels of journals,
workshops and conferences, faculty promotions and relevant tasks in ASNs. There are some
intrinsic problems that are involved in ranking of ASNs. These include the dependability of
the results of ranking with the attributes used as ranking criteria. Review of these ranking
criteria like the number of publications, the number of citations, the citation date, the context
of citations, the prestige of authors of citing article topic sensitivity, temporal dimension and
so forth would throw light on the role of these ranking criteria.
Various ranking methods have been proposed to quantify the scientific output and
quality of researchers/authors. Most of the research articles are co-authored by multiple
researchers. Using generic ranking models do not necessarily generate satisfactory results,
as these methods tend to treat all authors equally, whereas each author may have
contributed to a co-authored work highly differently. Instead of simply counting and
grouping the publications and citations of researchers, more appropriate and sophisticated
methods for author ranking are expected to produce far better results for decision making
and are thus much needed. Apart from ranking methods, several other areas are well
investigated for web databases, such as, research collaboration (Guns and Rousseau, 2014),
citation content analysis (Zhang et al., 2013), research community mining (Daud et al., 2009a;
Daud and Muhammad, 2012), citation recommendation (Daud et al., 2009).
Gupta et al. (2013) had published their results on a comparative study of various
link analysis methods. They discussed the pros and cons of several generic ranking
algorithms including citation count, PageRank, and Hyperlink-Induced Topic Search
(HITS). Jiang et al. (2013) published another comparative study on link analysis methods.
They studied link analysis methods and compared citation counting and summation of
paper ranks. Different from these two prior surveys, in our work we studied the author
ranking methods in a more thorough and detailed way –we came up with a classification
structure for the existing ranking methods and we incorporated more methods into our
investigation: probabilistic and learning-based methods. We expect our output
(as summarized in this article) to be more thorough and helpful to researchers who want
to hold a quick grasp of the status quo of the research in this topic area.
In this study, we classify a wide range of existing author ranking methods into three
main types based on their functionalities: these are link analysis, text similarity ranking and
learning-based methods. Each category is further divided into more specific subcategories.
The classification criteria utilized are inferred by thoroughly studying and analyzing the
available literature and contribute to the paper.
Currently, no benchmarking standard particularly designed for author ranking is
available. This makes the comparison process to be qualitative most of the time, instead of
quantitative. In this survey, we shall also discuss the limitations of the existing methods,
and in doing so, we put more emphasis on addressing future directions and inspiring new
ideas and methods for solving the current problems.
The article is structured as follows: following a general instruction (Section 1), in Section 2
we present some basic concepts which form a convenient basisfor the subsequent discussion;
in Section3, which reflectsour key contribution,we bring up a classification schemefor current
author ranking methods by which, we put existing ranking methods into three main types
(or categories) and numerous subtypes (or subcategories); in Section 4, we briefly address the
available data sets and evaluation metrics proposed forauthor ranking; in Section 5, we point
out future directions, and, finally, in Section 6, we conclude the paper.
2. Basic concepts
In this section, we introduce the basic concepts related to ASNs for the convenience of our
subsequent discussion.
98
LHT
36,1
2.1 Author and co-author
An author is an entity who can claim intellectual contribution in the accomplishment of the
research described in a scholarly article. The scholarly article can be published in the form
of a paper or a book and contribution can be based on author’s study, analysis and/or
experimentation. An academic author is usually a researcher conducting a study in a
particular academic discipline. When multiple researchers collaborate together and produce
a joint output in the form of a publication they are said to be co-authors of that publication.
2.2 References and citations
When an author cites or refers to an existing work in his/her paper, the cited work is called a
reference. (From now on, we may simply use the terms, paper or article, to imply any form of
publication produced by researcher). Because of the outgoing nature, the references
appeared in a paper are also called out-links. The references of a paper are cited in the text of
the paper and are listed at the end of the paper with details such as author names, paper
title, publishing venue and date, and page numbers, etc. The list of the references of a paper
is also known as the bibliography of the paper. Citation of a paper “A”occurs when the
authors of paper “B”mention a reference of paper “A”in paper “B”. In the context of ASNs,
we call each citation that a paper received (i.e. it is mentioned in another paper) an incoming
reference of the paper, which thus is also termed as an in-link.
2.3 Co-authorship networks, author-citation networks and paper-citation networks
The ASNs are usually represented as graphs in which nodes stand for authors or papers, and
edges represent a certain relationship between the nodes such as authorship ( between an
author node and a paper node) andco-authorship (between multiple authors with regard to a
commonpaper), author-citation (between twoauthor nodes via theirpapers), and paper-citation
(between two paper nodes). Accordingly, we differentiate three types of graphs (or networks):
co-authorship graph which represents a co-authorship network highlighting the collaboration
relationship among authors, author author-citation graph which represents a citation network
highlighting the citation relationship happened between authors, and paper-citation graph
which represents a citation network where papers directly refer to each other. Evidently,
citation relationship is a weaker relationship as compared to co-authors relationship, since
authors who cite each other’s work may not actually know each other, while authors who
collaborate on a common publication must (usually) have already known each other.
As examples, graph G1 in Figure1 illustrates co-authors relationship between authors (where,
A1 and A2 are co-authors of paper P1; P2 is solely authored by A2; P3 is co-authored by A1
and A3; and P4 is co-authored by A2 and A3); graph G2 is an author-citation graph (where, A1
cites A2 and A3; A2 cites A4; A3 cites A2; and A4 cities A1 and A3); and graph G3 is a
paper-citation graph (where, P1 cites P2 and P4; P2 cites P4; P3 cites P2; and P4 cites P3).
P1
P2
P3
P4
A1
A2
A3
A1
A3
A2
A4
P1
P4P2
P3
G1 G2 G3
Figure 1.
Examples of
co-authorship graph
(G1), author-citation
graph (G2), and
paper-citation
graph (G3)
99
Academic
social
networks
To continue reading
Request your trial