Fairness & Explainability in LLMs

Bias detection and semantic enrichment for large language models

Fairness & Explainability in LLMs

Overview

This project investigates fairness, scalability, and explainability in large language models.

Funding

  • Insight Research Ireland (Grant Number 12/RC/2289_P2)

Duration

Nov 2023 – Aug 2025

Team

  • PI: Dr Ihsan Ullah
  • Co-PIs: Edward Curry, Andy Donald, John McCrae

Researchers

Andy Donald; Apostolos Galanopoulos; Atul Kumar Ojha; Edward Curry; Emir Muñoz; Ihsan Ullah; John McCrae; Manan Kalra; Sagar Saxena; Talha Iqbal

Description

The project focuses on:

  • Automated generation of Model/Data/Dataspace schemas
  • NLP + RDF-based semantic enrichment
  • Bias detection using Transformer models:
    • DistilBERT, RoBERTa, ELECTRA, XLNet
  • Open data catalogue using CKAN & Airflow
  • FAIR AI documentation and transparency

Thematic Areas

  • Artificial Intelligence / Machine Learning
  • Data Science
  • Knowledge Graphs & Linked Data

Impact

  • SDG 9: Industry, Innovation & Infrastructure