Skip to content

shihabshahrier/BCS-Q-Classifier-agent

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

BCS-Q-Classifier

A Python tool to automatically classify Bangladesh Civil Service (BCS) preliminary exam questions into predefined categories using AI.

Features

  • Classifies BCS MCQs into 10 subject categories
  • Uses DeepSeek-V3 model for accurate classification
  • Processes questions in chunks with automatic retries
  • Saves progress periodically to prevent data loss
  • Provides detailed classification statistics

Categories

  • Bengali
  • English
  • Bangladesh Affairs
  • International Affairs
  • General Science & Tech
  • Computer & IT
  • Math Reasoning
  • Mental Ability
  • Ethics & Good Governance
  • Geography

Requirements

  • Python 3.x
  • Required packages listed in requirements.txt

Setup

  1. Clone the repository
  2. Install dependencies:
pip install -r requirements.txt
  1. Create a .env file with your Nebius API key:
NEBIUS_API_KEY=your_api_key_here

Usage

  1. Place your input CSV file containing BCS questions in the project directory
  2. Run the classifier:
python main.py

The script will:

  • Process questions in batches of 25
  • Save intermediate results every 5 chunks
  • Generate a final CSV with classifications
  • Display classification statistics upon completion

Input Format

The input CSV should contain a "question" column with BCS MCQs.

Output

The script generates:

  • bcs_with_categories.csv: Final output with classified questions
  • bcs_with_categories.csv.partial: Intermediate saves during processing

Configuration

Key parameters can be adjusted in main.py:

  • CHUNK_SIZE: Number of questions processed per batch (default: 25)
  • MAX_RETRIES: Number of API retry attempts (default: 3)
  • RETRY_DELAY: Seconds between retries (default: 2)

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages