Revisiting the uniformity and inconsistency of slow-cited papers in science. Issue 1 (February 2023)

• Record Type: Journal Article
• Authors: Miura, Takahiro; Asatani, Kimitaka; Sakata, Ichiro
• Is Part Of: Journal of informetrics. Volume 17:Issue 1(2023)
• Publication Date: 2023-02
• Subjects: Slow-cited -- Sleeping beauty -- Delayed recognition -- Citation analysis -- Scientometrics -- Science of science; Library statistics -- Periodicals; Information science -- Statistical methods -- Periodicals; Bibliometrics -- Periodicals; Bibliothèques -- Statistiques -- Périodiques; Sciences de l'information -- Méthodes statistiques -- Périodiques; Bibliométrie -- Périodiques; 020.727
• DOI: 10.1016/j.joi.2023.101378
• Languages: English
• ISSNs: 1751-1577

Abstract

Highlights: Slow-cited papers have long been discussed to reveal why outstanding discoveries remain unnoticed. The paper compares 11 slow-cited measures to identify uniform and inconsistent features of delayed recognition. Papers extracted by the measures have small overlaps, and the measures are grouped into four categories by their citation patterns. Slow-cited papers tend to be small-team, disruptive, combining fairly novel knowledge and gradually spreading broader fields. Abstract: Quantitative analyses on delayed recognition indicated by slow-cited papers, including delayed papers and durable papers, have long been discussed to reveal why outstanding discoveries remain unnoticed. However, these analyses include contradictory arguments, such as which combinations of knowledge, over-specialization, or transdisciplinary factors have led to undervaluation. We claim that this is because the indicators of delayed recognition are methodologically similar but capture conceptually different phenomena. Subsequently, this paper examined the overlap of 11 slow-cited measures to identify the uniformity and inconsistency of delayed recognition. Consequently, each measure practically obtained different papers as delayed recognition objectively classified into four groups by citation feature clustering, albeit based on similar concepts. Despite the ambiguity, we found that all delayed recognition measures extract papers that are more likely to be single-author projects that make disruptive contributions to more diverse fields without extremely novel nor conventional knowledge combinations that have been gradually awakened, compared to the null models. This result is robust when applying other hyperparameters, research topic-controlled null models, year-controlled null models, and other fields. This strongly indicates that delayed recognition leads to the reconstruction of a new direction of science and contributes to pioneering a revolutionary research topic. The source code for extracting slow-cited papers is available online. 1