Portfolio

NLP Projects Portfolio

Fake News Classifier

• Trained a model that classifies real and fake news stories on the test dataset with a high degree of accuracy
• BERT: 99% accuracy
• Logistic Regression: 98% accuracy
• Naive Bayes: 94% accuracy

Natural Language Understanding Website

• Translation
• Extractive Summarization
• Topic Modeling (Word Cloud)
• Sentiment Analysis
• Named Entity Recognition

Topic Modeling on Complete Corpus of Edgar Allan Poe’s Short Stories

• Created a model of the 6 main topics in a corpus of all 70 of Edgar Allan Poe’s short stories
• Displayed word clouds to show common words before and after text preprocessing
• Included a list of the 50 most frequently occurring nouns and adjectives in the corpus
• Used Python libraries such as gensim, pyLDAvis, and spaCy