Original scientific paper
https://doi.org/10.31534/engmod.2026.2.ri.02m
An Improved TF-IDF Method for Handling Keyword Extraction Issue: An Example of Biocatalysts Corpus
Liang-Ching Chen
; Department of Foreign Languages, R.O.C. Military Academy, Kaohsiung 830, TAIWAN
Sheng-Yu Wen
; Department of Chemistry, R.O.C. Military Academy, Kaohsiung 830, TAIWAN
Kuei-Hu Chang
; Department of Management Sciences, R.O.C. Military Academy, Kaohsiung 830, TAIWAN
*
Jeng-Fung Hung
; Graduate Institute of Science Education and Environmental Education, National Kaohsiung Normal University, Kaohsiung 824, TAIWAN
* Corresponding author.
Abstract
Keywords are essential for identifying key topics in high-impact research areas such as the use of biocatalyst in chemistry, biotechnology, and environmental science. Although term frequency-inverse document frequency (TF-IDF) is a fundamental natural language processing (NLP) and artificial intelligence (AI) technique, it faces limitations in capturing current trend information. Its reliance on individual document statistics often leads to ineffective keyword ranking during keyword extraction. To overcome these constraints, this study proposes an improved TF-IDF method and validates it using 15 highly cited articles indexed in the Web of Science (WOS) in 2023. The main contributions are: (1) developing a new NLP tool capable of extracting quantitative parameter data from specialized target corpora; (2) defining document importance through the journal impact factor (IF), a long-standing indicator of journal influence in WOS; (3) integrating TF, IDF, and IF into a novel equation derived from the risk priority number (RPN) framework; and (4) enabling keyword ranking across the entire corpus to identify prominent high-impact research themes represented in the selected literature.
Keywords
Hrčak ID:
353214
URI
Publication date:
10.11.2026.
Visits: 0 *