Impact factor based on logarithmic correction for papers' citations and studies on its category normalization.

The objective of this study is to address the disciplinary variations in Journal Impact Factor (JIF), enabling its reasonable application in the evaluation of journals across diverse disciplines. To achieve this, the study utilized journals indexed in the Journal Citation Reports and considered papers from eight distinct disciplines as the research subjects. The citations of papers within each academic field were transformed into normal distributions, each with a different base logarithm. Subsequently, the logarithm-corrected impact factor (IF_log) for each journal was calculated. Category normalization for the IF_log (cnIF_log) was conducted by dividing each journal's IF_log by the logarithmic correction aggregate impact factor of the discipline to which it belongs and the superiorities of cnIF_log in the evaluation of academic journals across different disciplines were demonstrated empirically. The results showed that taking different base logarithms could convert the citations of the journal papers into an approximately normal distribution. The IF_log of the eight disciplines' journals was normally distributed, and the IF_log1.5, IF_ln, IF_log5, and IF_log10 of the journals were 100% positively related (r=1.000, P=0.000) both in the same and different disciplines. Compared with the impact factor (IF2018), average impact factor percentile (aJIFP), journal index of PR8 (JIPR8), IF_log and relative impact factor (rIF2018), and other indicators, cnIF_log1.5 exhibits the least variation among the journals from the eight different fields. In comparison to aJIFP and JIPR8, it demonstrates the highest correlation, ideal discrimination, and stability. It is concluded that cnIF_log1.5 is an ideal journal evaluation indicator in same or different disciplines. [ABSTRACT FROM AUTHOR]