Sentiment Analysis of Public Comments on the YouTube Video “Trump Unveils Sweeping Global Tariffs in Watershed Moment for World Trade” by BBC News Using the Long Short-Term Memory Method

Abstract

This study aims to analyze public sentiment towards the announcement of global tariffs by the President of the United States, Donald Trump, using the Long Short-Term Memory (LSTM) method. The analysis focused on user comments from one video uploaded by BBC News on its official YouTube channel, titled “Trump Unveils Sweeping Global Tariffs in Watershed Moment for World Trade.”. Sentiment analysis is performed by classifying public comments into positive or negative sentiment categories, through preprocessing stages such as case folding, cleansing, normalization, stop words, stemming and tokenization. The processed data is then used to train and evaluate the LSTM model, which is known to capture temporal relationships and contextual meaning in text data. The results showed that the sentiment was negative, with 64.6% of the comments showing negative sentiment and only 34.4% showing positive sentiment. The performance of this LSTM method gives a performance of 76% Accuracy with 77% precision, 84% recall, and 81% f1-score on negative sentiment and 74% precision, 64% recall, and 69% f1-score on positive sentiment. These findings demonstrate the public's critical view of Donald Trump's global tariff policy and confirm the effectiveness of the LSTM method in extracting sentiment trends from online discussions. This research contributes to the analysis of public opinion in the context of international economic policy.