Figure 2
The code is written as follows: BEGIN PROCEDURE Import data processing libraries Initialize sentiment analysis engine Load Twitter dataset containing public discourse on the incident FOR EACH tweet in dataset: Extract textual content Analyze tweet for presence of 20 predefined sentiment categories: Double slash Categories include: Denial, Anger, Bargaining, Depression, et cetera. Double slash Sentiment classifications by querying a GPT-based N L P model with a system prompt and Top20Tags schema. Double slash Uses Open A I action model 'g p t-4.1-nano-2025-04-14'. Double slash system underscore prompt contains context and instructions for assigning the 20 tags. Double slash The provided context was: Double slash An incident report and research goal Double slash The instructions were to: Double slash Assign each of the 20 tags a boolean value (True or False) Double slash Return the result as a JSON object, strictly adhering to the Top 20 Tags schema Generate binary classification (present or absent) for each sentiment category Record sentiment classifications in the dataset Update analysis statistics END FOR Aggregate sentiment distribution across the entire dataset Slash asterisk The resulting data enables quantitative measurement of public reaction patterns following the security incident asterisk Slash END PROCEDUREPseudo code for the quantitative sentiment analysis. Source: Authors’ own work