Interactive Parliamentary Debate Adjudication System

Omkar Joshi*, Priya Pitre*
NeurIPS Human-in-loop-learning Workshop | New Orleans, LA (2022) | [Link]

Overview

This project proposes an interactive system that combines human expertise with machine learning to improve parliamentary debate adjudication.

Key Features

  • Human-in-the-loop learning framework
  • Argument structure analysis
  • Automated evaluation metrics
  • Interactive feedback mechanism

Approach

The system learns from human judges’ decisions while incorporating:

  • Natural language understanding of arguments
  • Debate-specific scoring heuristics
  • Iterative refinement through interaction
  • Comprehensive evaluation methodology

Technologies

Built using PyTorch, scikit-learn, and NLTK for NLP components.

← Back to Research