A comprehensive approach to preprocessing data for bibliometric analysis

Abstract

Abstract Bibliometric analysis, also known as bibliometrics, has been conducted for several decades to evaluate scientific research based on data available on bibliographic platforms, such as the popular Web of Science or Scopus. Research papers which include bibliometric analysis typically ignore the problem of bibliographic data preprocessing, in particular its important aspect—data cleaning. Discussion of bibliographic data preprocessing in the literature is sparse and scattered; studies usually address selected single components of the entire endeavour. This study aims to fill the gap as a review article, extensively analysing the problem, presenting issues arising from the structure of bibliographic data, combining data from various sources, creating thesauri and conducting bibliometric analyses, also through the author’s own experience. A brief description of the most popular software dedicated to bibliometrics, such as BibExcel, Bibliometrix, CiteSpace, CitNetExplorer, SciMAT, Sci2 Tool, and VOSviewer, is also provided, highlighting the operations available in these applications for the preliminary processing of bibliographic data. The work allows us to draw the following conclusions. The task is more difficult and demanding than some authors suggest or unclearly claim has already been accomplished, without providing additional details. Data cleaning operations are carried out at various stages of preprocessing, sometimes repetitively, and the order in which they are performed may be significant as it determines the success or failure of the process, in particular when combining data from different sources. There is no software which allows automatic execution of the entire preprocessing procedure of bibliographic data. Moreover, manual work is inevitable at various stages of the process. The contribution of this work to the field of bibliometric analysis is expressed in the form of a methodological synthesis, which involves the holistic consideration of the discussed issue, enabling a more comprehensive understanding of it.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 46 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.