Summarization & RAG Benchmarks
Reliable evaluation for domain-specific LLM tasks
Designing datasets and metrics for grounded summarization and retrieval-augmented generation.
Research Overview
We are a graduate-led research group advancing applied AI with an emphasis on NLP, LLMs, and human-centered, responsible methods. Founded in Summer 2024, we explore ideas from fundamentals to deployment.
Bayesian inference, graphical structures (BNs, HMMs), and uncertainty-aware modeling for real-world decision support.
Transformers, diffusion, and representation learning with a focus on robustness, efficiency, and deployment.
Our NLP work sits at the intersection of deep learning and computational linguistics: text
summarization, retrieval-augmented generation, and evaluation of LLMs for domain tasks. We emphasize
ethical practice—bias analysis, accessibility, and transparent reporting—alongside performance.
Prompt engineering, RAG, evaluation protocols, and alignment approaches for trustworthy, task-ready LLM applications.
Fairness, explainability, privacy, and reproducibility from data collection through deployment and monitoring.
Knowledge graphs and gaze-based attention modeling to connect meaning with human behavior signals.
Interactive systems, visual analytics, and graph mining at scale for insight and decision support.
Current Projects
Fairness & Interpretability Tooling
Actionable diagnostics for model behavior
Bias detection, counterfactual testing, and report templates for transparent deployments.
Knowledge Graphs with Eye-Tracking
Linking meaning with human attention
Gaze-informed semantic relations for better retrieval and user modeling.
Interactive Network Analytics
From data to decisions on the web
Graph mining, anomaly detection, and visual dashboards for large-scale systems.
Publications
See our latest papers and preprints, sorted newest → oldest.
View PublicationsHow to Join
We welcome motivated students from any major. New members complete a fundamentals tutorial, then choose a topic or join an existing team. We meet weekly to share progress and plan next steps.
Email Dr. Nafa to JoinResearch Group Policy
Students from all majors and class years are welcome. To join, email f.nafa@northeastern.edu.
We meet weekly at a consistent time. Regular attendance is expected; notify the faculty advisor if you must miss a session.
Propose a new idea or join an ongoing effort. All projects require faculty approval and periodic progress updates.
We pursue grants and internal opportunities. Funds support datasets, compute, travel, and dissemination (e.g., conferences).