Job Classification and Recommendation System
- Extracted and structured around 100k job descriptions by automating scraping on job boards using bs4 and requests,
- Tokenized job descriptions applying BERT transformer, decreasing corpus size by 30%, and vectorized into matrix with tf-idf
- Trained and Evaluated Naive Bayes, Gradient Boosting, Random Forest, Support Vector Machine and Logistic Regression to classify job vectors into job titles and attained 88% accuracy with Random Forest
- Recommended job titles by calculating cosine similarity between classified jobs and user resumes
