This repository contains an NLP-based analysis of President Donald J. Trump’s second-term speeches using NLTK, Scikit-Learn, and BERT Transformers. The project aims to extract key linguistic patterns, topic distributions, sentiment analysis, and named entity recognition from his speeches.

The source of the speeches was ‘The American Presidency Project’ maintained by the University of Santa Barbara. The speeches and addresses used were of the following categories:

1. Interviews
2. News Conferences
3. Spoken Addresses and Remarks
4. Farewell Addresses
5. Inaugural Addresses The pool reports were not included as they are mediated via reportage.

• Sentiment Analysis using BERT Transformer Model
• N-gram Analysis (Bigrams & Quadgrams)
• Topic Modeling using NMF (Non-Negative Matrix Factorization)
• Named Entity Recognition for frequently mentioned leaders, cabinet members, and countries
• Word Cloud Analysis of frequently used adjectives, nouns, and verbs
• Speech Engagement Analysis (Applause & Laughter frequency)
• Immigration-Related Word Cloud Analysis

• Majority of the sentences in the speeches have positive sentiment, with a significant number of negative statements as well.
• Sentiment Distribution: ‘Positive sentences’: 2921, ‘Negative sentences’: 1923, ‘Neutral’: 0

• Frequent bigrams include “thank much”, “united state”, “president well”, showcasing key phrases in his speeches.

• Extracted 5 key topics from the speeches, including themes on governance, national security, economy, and audience interaction.

• Four-word phrases reveal structured patterns in policy discussions and rally rhetoric. The motto of the Trump campaigh ‘Make America Great again’, as expected is represented as the most frequent quadgram.

• Shows where audience engagement is highest, analyzing “applause” and “laughter” frequency.
• Total words spoken in official speeches since inauguration were 23981: Interrupted by applause every 147 words and by laughter every 184 words

• Highlights Trump’s second-term cabinet picks with highest mentions. Elon Musk is mentioned most.

• America, China, Israel, Canada, Iran, Ukraine, Russia are the most frequently mentioned countries in speeches.

• Joe Biden, Benjamin Netanyahu, Vladimir Putin, Justin Trudeau lead the mentions in Trump’s speeches.

• Frequent words include “border, southern, invasion, security, wall”, indicating major speech themes.
• Remarkable that the word ‘immigrant’ makes no apprearence in any speech,whereas ‘illegal’, ‘alien’ and ‘immigration’ are found.

• Shows key entities (places, policies, events) focused on in speeches.

• Major policy themes extracted from speeches.

• Frequently used action words indicating key speech intentions.

• Python (for text processing & NLP)
• NLTK (Natural Language Toolkit)
• Scikit-Learn (TF-IDF & NMF topic modeling)
• Hugging Face Transformers (BERT Sentiment Analysis)
• Matplotlib & Seaborn (for visualizations)

1. Clone the repository:

   git clone https://github.com/SYEDFAIZAN1987/Trump-Second-Term-Speeches-NLP-Analysis.git

    
2. Install dependencies:

   pip install -r requirements.txt

    
3. Run the Jupyter Notebook:

   jupyter notebook "Trump Speeches NLP Analysis.ipynb"

    

For any questions or collaborations, feel free to reach out:

• Author: Syed Faizan
• Email: faizan.s@northeastern.edu
• GitHub: SYEDFAIZAN1987

This project is licensed under the MIT License. Feel free to use and modify it!