没有合适的资源?快使用搜索试试~ 我知道了~
Authors' status and the perceived quality of their work Measur
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
0 下载量 10 浏览量
2022-07-09
16:00:12
上传
评论
收藏 920KB PDF 举报
温馨提示
试读
11页
Authors' status and the perceived quality of their work Measuring citation sentiment change in nobel article
资源推荐
资源详情
资源评论
Authors’ Status and the Perceived Quality of Their Work:
Measuring Citation Sentiment Change in Nobel Articles
Erjia Yan
College of Computing and Informatics, Drexel University, Philadelphia, PA. E-mail: [email protected]
Zheng Chen
College of Computing and Informatics, Drexel University, Philadelphia, PA
Kai Li
College of Computing and Informatics, Drexel University, Philadelphia, PA
Prior research in status ordering has used numeric indi-
cators to examine the impact of a status change on the
perception of a scientist’s work. This study measures
the perception change directly as reflected in citation
sentiment, with the attainment of a Nobel Prize in Chem-
istry or a Nobel Prize in Physiology or Medicine consid-
ered the status change. The article identifies 12,393
citances to 25 Nobel articles in PubMed Central and
includes a control article set of 75 articles with 30,851
citances. The results show a moderate increase in cita-
tion sentiment toward Nobel articles postaward. Dynam-
ically, for Nobel articles there is a steady sentiment
increase, and a Nobel Prize seems to co-occur with this
trend. This trend, however, is not evident in the control
article set.
Introduction
Status ordering is a fundamental area of sociological
research. Humans are hardwired to understand their position
within a group, from children in playgrounds (Bothner,
Godart, & Lee, 2010), to pupils in classrooms (Patterson,
Kupersmidt, & Griesler, 1990), to employees in work-
places (George, Dahlander, Graffin, & Sim, 2016; Harley,
1999). This near-obsession with knowing one’s status and
advancing in the status ordering may be attributed to the
availability of resources: the scarcer the resources, the more
desirable they are to us. In this competition for resources,
higher status confers an advantage, thus making such status
more appealing.
In science, the disproportionate concentration of resources
is known as the Matthew Effect. Coined by Robert Merton
(1968), the term describes the phenomenon whereby emi-
nent scientists often get more scientific credit compared
with those who are relatively unknown. Thus, a change in
status will inevitably result in a marked change in one’s
access to resources, social capital, and self-assurance, among
other characteristics. Past research has established the exis-
tence of the Matthew Effect in a variety of scientificand
scholarly settings, including scientific collaboration (Barabâsi
et al., 2002; Newman, 2001a, 2001b, 2001c, 2004), informa-
tion networks (Clauset, Shalizi, & Newman, 2009), distribu-
tion of citations (Bensman, 2008; Redner, 1998; Seglen,
1992; Stringer, Sales-Pardo, & Amaral, 2010), and word fre-
quency distributions (Newman, 2005), among many other
types of empirical data (Perc, 2014).
Prior studies have largely relied on numeric indicators
to approximate resources when examining status ordering.
The number of collaborators, for instance, has been used
as a proxy for social capital (Barabâsi et al., 2002; Newman,
2001a, 2001b, 2001c, 2004), and number of citations has
stood in for the concept of scientific rewards (Bensman,
2008; Redner, 1998; S eglen, 1992 ; Stringer et al., 2 010).
Although these studies have provided quantitative evidence
of the Matthew Effect, we have made little headway in
understanding the impact of status change on subjective per-
ceptions of one’s work. This lack of progress is likely due to
the lack of proper instruments with which to measure percep-
tions. In the present study, we aim to address this problem
by tackling the complexity of citation sentiments, as used to
measure the impact of authors’ status change on how others
cite their work. Among all the possible status changes in a
scientist’s career, winning a Nobel Prize is unparalleled.
Thus, this study uses Nobel Prizes as the status change and
Received April 3, 2018; revised March 13, 2019; accepted March
21, 2019
© 2019 ASIS&T • Published online Month 00, 2019 in Wiley Online
Library (wileyonlinelibrary.com). DOI: 10.1002/asi.24237
JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY, 00(0):1–11, 2019
measures the citation sentiment of articles citing a Nobel arti-
cle before and after its author’s investiture.
As the most highly esteemed science award, the Nobel
Prize has been studied in the context of scholarly commu-
nication from several perspectives, as have its laureates.
Topics of such research include prize prediction (Ashton &
Oppenheim, 1978; Garfield, 1986; Gingras & Wallace, 2010),
sources of funding of Nobel articles (Tatsioni, Vavva, &
Ioannidis, 2010), Nobel Prize effects on citation impact
(Farys & Wolbring, 2017; Frandsen & Nicolaisen, 2013), pat-
terns of collaboration (Chan, Önder, & Torgler, 2015), collab-
oration networks (Wagner, Horlings, Whetsell, Mattsson, &
Nordqvist, 2015), and the likelihood of laureates to obtain
more awards after their Nobel Prize (Chan, Gleeson, &
Torgler, 2014).
Because of the Nobel Prize ’s unmatched prestige, schol-
arly communication researchers have been curious to find
out whether Nobel laureates behave differently from aver-
age scientists, and whether their characteristics can be identi-
fied and used to predict future laureates. It has been found
that although Nobel laureates differ from average scientists
in their citation impact, bibliometric indicators alone are not
able to predict prize winners; rather, such indicators simply
identify a group of elite scientists (Ashton & Oppenheim,
1978; Garfield, 1986; Gingras & Wallace, 2010). In addi-
tion, there seems to be a cascading effect: a Nobel Prize
affects not only the citation of the laureate’swork,butalso
that of its cited references, as shown in a case study of
Robert J. Aumann (Frandsen & Nicolaisen, 2013). How-
ever, when citations are normalized based on the articles’
publication year, s uch an effect is no longer observed
(Farys & Wolbring, 2017). Apart from citation impact,
studies have also shown that Nobel laureates have fewer
coauthors compared with a control group and tend to pos-
sess higher social capital, by bridging different communi-
ties of researchers (Wagner et al., 2015).
Particularly relevant to the scope of this research are the
studies examining pattern changes following prize recogni-
tion. Patterns examined to date include collaboration patterns
(Chan et al., 2015), citation profiles (Azoulay, Stuart, &
Wang, 2013), and the ability to obtain awards (Chan e t al.,
2014). Studies have found that Nobel laureates form fewer
collaborations with new coautho rs postaward tha n preaward
(Chan e t al., 2015), and their likelihood to obtain awards
drops after they win a Nobel Prize (Chan et al., 2014). In a
similar vein, a study showed a postaward citation increase
to articles published before their authors earned a Howard
Hughes Medical Institute (HHMI) Investigator recognition
(Azoulay et al., 2013). This confirms the theory that people
change their perceptions of the quality of others’ work when
there is a shock to the authors’ status (such as a highly
esteemed award).
This study builds on prior work by using citation senti-
ment to directly measure the change in perception of a sci-
entific article after a change in the author ’ s status. The
change in this case is the receipt of either of two Nobel
Prizes: the Nobel Prize in Chemistry or the Nobel Prize in
Physiology or Medicine. The set of citation sentences (that
is, “citances”) from citing articles in PubMed Central (PMC)
to Nobel laureates’ work is extracted and processed for senti-
ment analysis. We compare the sentiment of citances from
before and after the work’s author won a Nobel Prize and
use a set of control group articles to benchmark the senti-
ment change. This approach allows us to address one central
research question in this article: to what extent does a change
in authors’ status—such as winning a Nobel Prize—change
the perception of the quality of their work as measured by
citation sentiment?
Literature Review
Citation analysis is strongly based on the relationship
between citing and cited documents. The classic explana-
tion of citation behavior is the normative theory, which
regards citations as a means of paying an intellectual debt
to the authors being cited (Gar
field & Merton, 1979; Kaplan,
1965). Despite its importance in the hi story of citation
analysis, subsequent generations of researchers increas-
ingly questioned t he normative assumptions behind this
theory (for example, Bornmann & Daniel, 2008) and
accepted a new approach to citation data that focuses on
the content and context of citation relation.
This newer program of research was officially named
content and context analysis in the early 1980s (Small,
1982), defined as including those studies examining the
“particular message or statement within the citing docu-
ment containing the reference” (p. 288). This type of analy-
sis can be traced back to Lipetz’s pioneering work (1965),
in which the author identified four categories of relationship
between the document pair: original scientific contribution
of citing article, contribution other than original scientific
contribution, identity or continuity relationship between arti-
cles, and disposition of the scientific contribution of the
cited article to the citing article. A large number of studies
under this category have been conducted since then
(Case & Higgins, 2000; Chubin & Moitra, 1975; Duncan,
1981; Frost, 1979; McCain & Salvucci, 2006; Moravcsik &
Murugesan, 1975; Spiegel-Rösing, 1977), most of which
developed a classification scheme based on a set of scien-
tificdocuments.
Cronin (1984) correctly observed that most of these
studies fail to form a “cumulative endeavor” (p. 35), despite
the fact that some regularities are evident in these studies.
His observation is supported by work which identifies and
summarizes the facets underlying these classification
schemes. An article by Zhang, Ding, and Milojevi
c (2013),
for example, identifies six principles embedded in all these
studies, including the type of motivation, level of impor-
tance, type of resource, function of citing, type of dis-
position/sentiment, and location of mentioning (p. 21).
Despite its status as one of th e central principles of con-
tent and context analysis, the sentiment polarity expressed
in texts is a relatively late-coming research topic. Although
sentiment polarity has been explored by researchers from
2 JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2019
DOI: 10.1002/asi
the perspectives of citation analysis (for example,
MacRoberts & MacRoberts, 1984; Ziman, 1968) and com-
putational linguistics (for example, Wilks & Bien, 1983), it
did not become a popular topic in its own right until the
beginning of the 20th century, a phenomenon partly driven
by the rise of machine-learning methods and the increasing
availability of data sets (Pang & Lee, 2008).
As a response to these new developments in computa-
tional data science, data-driven sentiment analysis was soon
introduced to the field of citation analysis. Awais Athar and
colleagues discussed the difficulties of applying sentiment
analysis to citation statements and proposed a support vector
machine (SVM) method based on different aspects of sen-
tence structure (for example, Athar, 2011, 2014; Athar &
Teufel, 2012a, 2012b). Machine-learning techniques are a
dominant methodological element in citation sentiment stud-
ies. In addition to SVM (see also Hernández-Alvarez &
Gómez, 2015; Kim & Thoma, 2015; Xu, Martin, &
Mahidadia, 2013), notable examples include random
forest (Abu-Jbara, Ezra, & Radev, 2013; Parthasarathy &
Tomar, 2014), naïve Bayes (Butt et al., 2015; Sula &
Miller, 2014), and neural network methods (Lauscher,
Glavaš, Ponzetto, & Eckert, 2017). Within this category,
SentiWordNet, a lexical resource for opinion mining that
is partly based on a semisupervised machine-learning method,
has also been used in a number of studies (Goodarzi,
Mahmoudi, & Zamani, 2014; Sendhilkumar, Elakkiya, &
Mahalakshmi, 2013).
It should be noted that most of the above-mentioned
studies focus primarily on establishing and testing a meth-
odological framework for citation sentiment analysis. With
more citation analyses taking the sentiment into consider-
ation, we are also gaining more insights about how social
factors help to construct scientific knowledge. For example,
before the advancement of computational citation semantic
analysis, Small (2011) concluded that there is a correlation
between prominent sentiments and competing knowledge
claims. On the other hand, a more recent study (Ma, Nam, &
Weihe, 2016) suggested that an author’s reputation is a
reliable predictor of the sentiment toward one’s works.
Although these findings contributed to a deeper under-
standing of citations, we are still far from fully understand-
ing the relationship between citation sentiment and the
authors who receive it. Thus, in this artic le we examine the
impact of a change to authors’ status on the perception of
their work as measured by citation sentiment.
Data
Two sets of articles are distinguished, and both are
introduced in this section. The first set is the so-called
“Nobe l articles,” written by Nobel Prize laureates and thought
to be closely connected with the conferral of their respective
Nobel Prizes. The Advanced Information section of each
award at www.nobelprize.org served as our data source for
identifying Nobel articles. Wefocusedontwoawards,the
Nobel Prize in Chemistry and the Nobel Prize in Physiology
or Medicine, because articles that cite these Nobel articles are
more likely to be included in PubMed Central (PMC)—a
large full-text repository freely accessible to the public. Nobel
articles, as we define them, are typically referred to as “land-
mark” articles in the Advanced Information section or were
published within a 3-year window of the landmark articles by
thesameteamsofauthors.
Most articles in PMC were published after 2008; thus,
to monitor the impact of an award on citation sentiment
change, we narrowed Nobel Prizes to those conferred between
2010 and 2015. This time frame provides a minimum 3-year
citation window for the oldest (2008) article in PMC to cite
a Nobel article at the time of data collection. To effectively
link a Nobel article with its citing articles, the Nobel article
should be indexed in PubMed and have a PubMed ID
(PMID), but it do es not itself need to be included in PMC.
After removing those Nobel articles that lack a PMID and
those with fewer than 20 citations before their respective
Nobel Prize, we included in the final data set 14 articles for
Physiology or Medicine (hereafter Medicine) and 11 for
Chemistry. The 2018 October version of PMC was used for
data collection. In all, these articles received 12,393 citances,
of which 4,677 took place before the award. The data set
used in this study is uploaded to Figshare (Yan, 2019).
The second set contains control-group articles with which
to benchmark the sentiment change in the Nobel articles. For
each Nobel article, we selected three control articles that each
had a similar citation count to the Nobel article and were
published within 2 years of the Nobel articles. These selec-
tion criteria ensure that the control articles a re sufficiently
similar to the Nobel articles that the major distinguishing fac-
tor is the Nobel Prize. In total, there are 75 articles in the
control group, receiving a total of 30,851 citances.
To calculate the number of citations an article has received
in PMC, one can count the number of times that the cited
article’s PMID occurred in other articles’ cited reference
sections. A full-text search does not need be conducted;
thus, we refer to this process as citation metadata matching.
Tracking down citances within an article’sfulltext,how-
ever, is less straightforward and presents several challenges.
The first challenge is to identify the exact sentence where
the citance appears and extract all words that surround the
citance in order to evaluate sentiment. To achieve this, for
every paragraph in the abstract and the document body, the
XML nodes for citations (<xref> nodes) were first replaced
by a special token, and then the full paragraph text was
tokenized and further split by the recognized punctuations.
Initially, when only considering the standard XML nodes
(<xref> nodes), for our Nobel article set, this returned less
than half of all citations Nobel articles received in PMC as
returned by a metadata matching.
Thus, we need to deal with the second challenge, which
is the lack of consistency in XML citation coding. We pro-
vide some examples here.
1. Different encodings are used for the dash character ‘-’
inside a multicitation instance such as “[11–17].” It may be
JOURNAL OF THE ASSOCIATION FOR INFORMATION SCIENCE AND TECHNOLOGY—Month 2019
DOI: 10.1002/asi
3
剩余10页未读,继续阅读
资源评论
博士僧小星
- 粉丝: 1924
- 资源: 5885
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- 基于Java的在线教育平台设计源码
- 升降压斩波电路电路图,buck-boost电路图
- Python 程序语言设计模式思路-结构型模式:组合模式:将对象组合成树形结构
- 毕业设计基于python矩阵分解的推荐算法研究源码+详细文档+全部数据资料 高分项目.zip
- 基于网络的入侵检测系统源码+数据集+详细文档(高分毕业设计).zip
- 微信小程序源码 旅行故事分享 - 面包旅行App界面设计与文本展示资源下载
- 微信小程序源码 创意互动游戏 - 你画我猜App下载
- 摸底考试_学生版20230305.py
- 课程设计基于FPGA数字钟课程设计源码+课设报告(95分以上).zip
- 基于Java的企业家申报系统设计源码
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功