Resume Ranking System

Overview

Part-1: sample.ipynb or directly use "get_similarity_score.py" file under scripts

Requirements

• Python 3
• cohere
• qdrant-client
• yaml
• json
• logging

Setup

1. Install the required packages using pip:

2. Ensure you have a valid API key for the Cohere API and Qdrant.

3. Update the configuration file scripts/similarity/config.yaml with your Cohere and Qdrant API keys.

Usage

1. Provide the paths to the JSON files containing keywords extracted from resumes and job descriptions in the READ_RESUME_FROM and READ_JOB_DESCRIPTION_FROM variables, respectively.

2. Run the app_similarity_score.py file.

Description

• The Python script that calculates the similarity score between a job description and a resume using embeddings generated by the Cohere API and the Qdrant vector database.
• config.yaml: Configuration file containing API keys for the Cohere API and Qdrant.
• Data/Processed/Resumes: Directory containing JSON files with keywords extracted from resumes.
• Data/Processed/JobDescription: Directory containing JSON files with keywords extracted from job descriptions.
• app_similarity_score.log: Log file containing information about the execution of the program.

Part-2: transformer_sample.ipynb

This project is a simple Python program that ranks resumes against a job description provided in a PDF format. It utilizes the sentence-transformers library for text encoding and similarity calculation, and PyPDF2 for reading PDF documents.

Requirements

• Python 3
• sentence-transformers
• PyPDF2

Install the required packages using pip:

Usage

1. Place the resume PDF files in the directory specified by the org_docs list.
2. Provide the path to the target job description PDF file in the target_document variable.
3. Run the script.

Description

• The Python script that extracts text from PDF resumes and a job description file, encodes them into embeddings using Sentence Transformers, calculates the cosine similarity between the job description and each resume, and ranks the resumes based on their similarity scores.

File Structure

• Data/Resumes: Directory containing PDF files of resumes.
• Data/JobDescription: Directory containing the target job description PDF file.

License

This project is licensed under the MIT License - see the LICENSE file for details.