YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
NLP_Resumes2Roles
Natural Language Processing project that creates a list of job roles that will match with a given resume. 4 Vectorization methods are implemented:
- TF-IDF
- Word2Vec
- GloVe
- Sentence BERT
Authors:
- Ethan Eckmann
- Neha Indolikar
- Eddie Lin
- Chi-Heng (Henry) Wei
Class:
- I320D: Text Mining & Natural Language Processing Essentials
- Dr. Abhijit Mishra
Demo: This code creates a UI that allows you to enter in the text of a resume and get the best matching job roles. There are customizable for the number of results and chosen vectorization method.
Analysis: This code was used for testing and developing the project. It contains a BERTScore section for evaluating Recall, Precision, and F1-score for all 4 vectorization methods.
Notes on running .ipynb files: The first time you run either the demonstration or the main analysis .ipynb file, it will take a very long time. This is because all the vectors/embeddings need to be created on the first run. Unfortunately, since GitHub has a max file size of 100MB these supporting files could not be included. The SentenceBERT embeddings take an extremely long time to generate if not using a GPU.