An Improved Corpus-Based NLP Method for Facilitating Keyword Extraction: An Example of the COVID-19 Vaccine Hesitancy Corpus

In the current COVID-19 post-pandemic era, vaccine hesitancy is hindering herd immunity generated by widespread vaccination. It critical to identify factors that may cause hesitancy, enabling relevant authorities propose appropriate interventions for mitigating such a phenomenon. Keyword extraction, sub-field of natural language processing (NLP) applications, plays vital role in modern medical informatics. When traditional corpus-based NLP methods are used conduct keyword they only consider word’s log-likelihood value determine whether it keyword, which leaves room concerns about efficiency and accuracy this extraction technique. These include fact method unable (1) optimize list machine-based approach, (2) effectively evaluate keyword’s importance level, (3) integrate variables data clustering. Thus, address aforementioned issues, study integrated word removal technique, i10-index, importance–performance analysis (IPA) technique develop an improved facilitating extraction. The top 200 most-cited Science Citation Index (SCI) research articles discussing were adopted as target corpus verification. results showed keywords Quadrant I (n = 98) reached highest lexical coverage (9.81%), indicating proposed successfully identified extracted most important from corpus, thus achieving more domain-oriented accurate results.