• Không có kết quả nào được tìm thấy

Altmetrics for Digital Libraries

N/A
N/A
Nguyễn Gia Hào

Academic year: 2023

Chia sẻ "Altmetrics for Digital Libraries"

Copied!
292
0
0

Loading.... (view fulltext now)

Văn bản

Basierend auf der Berichterstattung zeigen wir, dass Altmetrics-Daten in diesen Disziplinen spärlich sind, und wenn Altmetrics-Daten für reale Anwendungen (z. B. in Bibliotheken) betrachtet werden, können höhere Aggregationsebenen, wie z. B. Journalebene, ihre Spärlichkeit durchaus ausgleichen. Auf dieser Grundlage zeigen wir, dass altmetrische Daten in diesen Disziplinen spärlich sind und für reale Anwendungen (z. B. in Bibliotheken) höhere Aggregationsebenen, wie z B. auf Zeitschriftenebene, könnte den Mangel an altmetrischen Daten durchaus überwinden.

Motivation

A study by Raamkumar et al. (2018) examined whether tweet sentiments for specific computer science articles can tell about the performance of those articles (i.e., article quality – citations). In addition, altmetric information for a specific DOI can be accessed without any additional account or login, just by using a bookmark (Trueger et al., 2015).

Dissertation scope

Based on previous research findings, the authors found good coverage of the top 30 journals and their articles in these disciplines, with 77.5% of articles found in Mendeley and 38% in Altmetric.com. The coverage of a large number of journals and their articles is important in this research because from the number of journals/articles found with altmetric information we can discuss the presentation of altmetric information for different ranked journals.

Table 1.1: Handelsblatt ranking journals in E and BS and their coverages in each class
Table 1.1: Handelsblatt ranking journals in E and BS and their coverages in each class

Proof-of-concept

The first feature will present an analytical approach that will explore altmetrics for a large volume of E and BS journals/articles. It will highlight the methodology that will be used to derive altmetrics, particularly for journal articles in E and BS.

EconBiz portal

Currently, the relevance ranking methodology in EconBiz is based on matches in title, author, abstract and position and frequency of the search term in the article (EconBiz, 2012). One of the experiments that LibRank considered is ranking search results based on popularity factors (ie, citations;.

Figure 1.4: Screenshot of the EconBiz portal start page.
Figure 1.4: Screenshot of the EconBiz portal start page.

ZBW personas

In this case, Dorothee will select journals, e.g. based on journal articles mentioned in blogs. Based on chapter 5, we will identify social media sources in which economic literature is mostly found.

Research questions

Since Altmetric.com tracks 19 different attention sources for journal articles (see Chapter 3 for more details), this study examines all cited sources for journal articles in economics and business studies, and presents the top five sources with the highest scores (Chapter 5). Given the findings of this part of the study, we will highlight different types of academic statuses of Mendeley users who read articles cited in Altmetric.com attention sources (Chapter 5).

Scientific contribution

Considering the aforementioned scientific contribution of this thesis, we can therefore propose several forms of presentation of altmetric information as a proof of concept in library portals, especially in EconBiz with the aim of helping both researchers and libraries to identify relevant articles (e.g.

Published work

Enriching the knowledge of altmetrics studies by examining social media metrics for economics and business studies journals. We present a case study of articles published in 30 economics and business studies (EBS) journals using social media statistics from Altmetric.com.

Structure of the thesis

Investigating altmetric information for the top 1000 journals in the Handelsblatt ranking of economic and business studies. The results of this study show that altmetric information is significantly better present in articles published between 2016 and 2017.

  • Journals
    • The history of journals in economics and business studies
    • Indexing data sources for scientific journals
  • Citation analysis
    • Citation indexes as sources for information retrieval (IR)
    • Identifying “trendy topics” using citations
    • Citation behavior in economics
  • Journal level indicators as relevance filters
    • Journal Impact Factor as a source for filtering relevant journals
    • h-index as a source for filtering relevant journals
    • Journal rankings as a source for filtering relevant journals
  • Discussion

In addition, this chapter will highlight the research questions, dissertation structure, and published work. This chapter is entirely devoted to magazines, what they are, who uses them and for what purpose.

Scholarly use of social media

Classification of altmetrics

Mendeley and Altmetric.com

  • Mendeley, a social reference management system
  • Mendeley readership information
  • Altmetric.com
  • Altmetric attention score
  • Altmetric attention sources

Altmetric as sources for filtering highly cited articles

Altmetrics as sources for filtering “trendy topics”

Altmetric studies for economic literature

Journal level altmetrics

Altmetric challenges

Discussion

Journal selection strategy

Top 1,000 journals from economics and business studies

Data collection phase

  • Crawling Crossref as data service for E and BS journals
  • Datatable design for Crossref in MySQL
  • Crawling Mendeley for E and BS journals
  • Downloading information from Altmetric.com

Conclusion

Dataset I: Crossref coverage for journals in E and BS

Dataset II: Mendeley coverage for journals in E and BS

  • Mendeley readership information: reader counts
  • Mendeley readership information: discipline
  • Mendeley readership information: academic status
  • Mendeley readership information: country

Dataset III: Altmetric.com coverage for journals in E and BS

  • Altmetric.com—altmetric attention score
  • Altmetric.com attention sources
  • The impact of the top 5 Altmetric.com attention sources
  • Correlation of citation counts with all altmetric attention sources
  • Identifying trendy topics using altmetrics (proof of concept with latent dirichlet

Discussion and conclusion

The setup of the questionnaire

  • First part: The demographic information
  • Second part: The task assigned to the participants and related questions
  • Third part: How useful are the metrics

Survey dissemination

Results…

  • The demographics of the participants
  • Findings: Article selection based on the given task
  • Findings: Bibliometrics indicators and journal rankings
  • Findings: Altmetric information on journal level
  • Findings: General opinions on the usefulness of indicators
  • Findings: Open questions

Discussion and conclusion

Lessons learned from the main findings

Within this thesis, we first investigated the top 1000 E and BS journals (more than half a million articles) in two altmetric providers: Mendeley and Altmetric.com. At Altmetric.com, 91.3% of E and BS journals are found, and we detected moderate article shares (about 44%) for publication years 2011–2018. 132 The article publication year “2018” for our study describes the most recently published articles in E and BS since the altmetric data for the articles was obtained in early 2019.

Based on the insights gleaned from this research, we concluded that E and BS journals generally have similar Mendeley user patterns independent of the journals' position in the Handelsblatt ranking. Nineteen different Altmetric Attention sources are identified while exploring Altmetric.com data for 1,000 journals in E and BS (see Table 5.12). We examined the DOIs for E and BS found in Altmetric.com that accumulated attention in each of the sources.

By studying the top 1,000 journals for altmetrics, we learned that Altmetric.com's most prominent sources for articles in E and BS journals are Mendeley, Twitter, News, Facebook, Blogs, and Policy Documents. Mendeley (followed in this case by Altmetric.com) is the source that provides the most altmetric counts for E and BS articles. We selected only those articles from the top 1,000 journals in E and BS that are only mentioned in one particular social media source that Altmetric.com tracks.

Implication of this thesis

According to the coverage of altmetric information for E and BS journals (Chapter 5), a suggestion for presenting altmetric data in real-world applications (eg, in libraries) would be to consider higher levels of aggregation, such as on a daily basis. For example, Figure 7.1 shows the journal-level altmetrics for the article title “Evaluation, evaluation, and definitions of research impact: a review” published in the journal Research Evaluation. When introducing journal-level altmetrics, based on data from Altmetric.com,134 libraries that adopt and researchers developing such metrics should also consider introducing the use of coverage percentages, for example, the percentage of articles with an AAS in a journal, or average citations per article.

For example, a useful solution to represent the attention of articles despite the journal-level information is to highlight the percentage of articles in an AAS in a journal similar to those in Altmetric.com and ImpactStory. Journal-level altmetrics can also be used as indicators to filter journals that are popular on social media or to identify journals that are highly discussed online. Presentation of altmetric information at a journal level will support the second persona use-case (ie interested in journal ranking).

One possible way to implement and adopt altmetrics at the journal level is an application called "Journal Map" currently being developed by ZBW. In addition to journal-level surrogate metrics, article-level surrogate metrics can also play an important role for libraries (see Figure 7.1; implemented in EconBiz for some articles). Article-level or journal-level altmetrics may be limited to sources most used by economists for articles in E and BS journals.

Figure 7.1 Suggested  altmetric information  on journal level in  EconBiz. In the top 5% of  all articles found within this  journal
Figure 7.1 Suggested altmetric information on journal level in EconBiz. In the top 5% of all articles found within this journal

Limitations of the findings

Alternatively, if we consider only displaying a small set of trending and recently published articles, the library portal can provide an additional tab of information highlighted as Trending in Altmetric/Mendeley, describing only articles that have been published recently (eg, in July 2020) and have received the highest online attention compared to other articles in that dataset. For example, when the portal user will click on the "Trending on Altmetric" tab, this click will trigger the display event of the top 3 articles with the highest Altmetric score in E and BS disciplines published in the previous 5 months. Alternatively, according to the proof of concept (see Chapter 5), there is an opportunity to identify trending topics for articles in E and BS, especially those that have been published recently, and show the topics that have received the most attention online based on the dataset.

Additionally, researchers should be educated on the inherent limitations of each altmetrics provider and consider whether these indicators are fit for their intended purpose. The implications listed under 3 of the “Top 5 Altmetric Sources of Attention” should only be used in libraries with an economics focus. For example, articles published in psychology and clinical disciplines have more F1000 and News citations (Htoo & Na, 2017) and these sources should be used as primary sources instead.

171 Altmetric information is obtained and analyzed for articles published only in scientific journals. In addition, the data that Mendeley receives is based only on users who practice Mendeley. Therefore, a larger group of economists should be involved to draw an accurate statement considering bachelor students as well.

Future work

In addition to DOIs, preprints, in economics, known as "working papers" have other identifiers (eg, handles), and different levels of altmetrics coverage can be identified for these types of articles (Nuredini & Peters, 2019). The altmetric information suggested in this thesis depends on the longevity of two altmetric providers. 2020) asserted that the presence of altmetrics is clearly growing, but the authors are encouraging researchers to share their articles within social media, which will drive the development and application of altmetrics. The statement that economists are not familiar with most altmetrics is based on some academic statuses that this survey has drawn from the mailing list, which bachelor students, however, are not covered.

The use of altmetrics for ranking search results in library systems (eg EconBiz) and identifying trending articles should be evaluated with real users so that more precise conclusions can be drawn from the evaluations. A/B testing is a valuable method that presents two or more variants of the system to the users and determines which of the variations had a better performance (Dixon et al., 2011). The correct immediacy of altmetrics can play an important role as we can more precisely identify the early impact indicators and thus use this precise insight as a benefit component (eg how often should libraries update altmetrics for E and BS journals) for future implications in library systems.

Estimating the different level of accuracy of impact measurement as a function of citation window length. Validity of altmetrics data for measuring social impact: A study using data from Altmetric and F1000Prime. An extensive analysis of the presence of altmetric data for Web of Science publications across subject areas and research topics.

Top 500 Handelsblatt ranking journals in business studies, their classes, and

Top 500 Handelsblatt ranking journals in economics, their classes, and altmetric

Correlations of altmetric attention sources with citations on journal level

Economic and business studies topics using latent dirichlet allocation

Questionnaire

Survey open questions responses

Hình ảnh

Figure  1.2:  Exponential  growth  of  WoS  articles  by  year  across  all  disciplines
Figure 2.1: E and BS journal coverage in different data sources. Source: Clermont & Dyckhoff  (2012, p.5)
Figure 2.2: Trendy topics extracted from computer science articles published between 1990  and 2004
Figure 2.3: Top 5 economic journals and their citation behavior. Source: Kapeller et al
+7

Tài liệu tham khảo

Tài liệu liên quan

Chapters 1 through 11 introduce the basic concepts of financial management: return and risk, time value of money, financial accounting (including the impact of inflation on