How muhammadaliasgher/job-recommender-nlp works
Project type and purpose
• This is an NLP-powered recommendation engine delivered as a Jupyter Notebook. It implements an end-to-end data-science pipeline: data ingestion → preprocessing → embedding → similarity matching → evaluation → visualization.
• Purpose: Semantically match 200 job postings with 200 resumes using Sentence-BERT embeddings and cosine similarity.
File structure
• Only one executable artifact (job-recommender-nlp.ipynb) plus documentation (README.md) and license.
• No separate “frontend” or “backend” code—everything lives in a single notebook as sequential cells.
• Dependencies and model downloads are specified in README.
README for additional insights
• Key libraries: spaCy for preprocessing, sentence-transformers for embeddings, scikit-learn for similarity, matplotlib/seaborn for plots.
• Data sources are CSV files from Kaggle (job postings and resumes).
Detailed guidance
• “Evaluation” (manual + average similarity metrics)
• “Visualization” (bar plots, heatmaps)