Application of entity linking to identify research fronts and trends

Mauricio Marrone*

*Corresponding author for this work

Research output: Contribution to journalArticle

5 Downloads (Pure)

Abstract

Studying research fronts enables researchers to understand how their academic fields emerged, how they are currently developing and their changes over time. While topic modelling tools help discover themes in documents, they employ a “bag-of-words” approach and require researchers to manually label categories, specify the number of topics a priori, and make assumptions about word distributions in documents. This paper proposes an alternative approach based on entity linking, which links word strings to entities from a knowledge base, to help solve issues associated with “bag-of-words” approaches by automatically identifying topics based on entity mentions. To study topic trends and popularity, we use four indicators—Mann–Kendall’s test, Sen’s slope analysis, z-score values and Kleinberg’s burst detection algorithm. The combination of these indicators helps us understand which topics are particularly active (“hot” topics), which are decreasing (“cold” topics or past “bursty” topics) and which are maturely developed. We apply the approach and indicators to the fields of Information Science and Accounting.

Original languageEnglish
Pages (from-to)357-379
Number of pages23
JournalScientometrics
Volume122
Issue number1
Early online date1 Nov 2019
DOIs
Publication statusPublished - 1 Jan 2020

    Fingerprint

Bibliographical note

Copyright the Author(s) 2019. Version archived for private and non-commercial use with the permission of the author/s and according to publisher conditions. For further rights please contact the publisher.

Keywords

  • Burstiness
  • Content Analysis and Indexing
  • Entity annotation
  • Information Storage and Retrieval
  • Natural Language Processing
  • Text analysis

Cite this