Lorenzo Molfetta

I'm a PhD student in Computer Science at the University of Bologna, where I am working on Natural Language Processing and Information Retrieval. My research interests focus on developing end-to-end explainable generative systems for question-answering, particularly in the biomedical and legal domains.

I work in the UniboNLP research group under the supervision of Professor Gianluca Moro, where we focus on innovative approaches to retrieve and inject relevant information to enhance the performance and reasoning capabilities of Large Language Models. Recently, we have been exploring methods to extract structured information graphs that can be injected into language models to improve explainability while grounding critical information and avoiding hallucinations.

My ongoing research works are focusing on:

  • Retrieval systems for Large Action Models

  • Memory-augmented Language Models

  • Improving reasoning capabilities of LLMs through end-to-end token injection mechanisms

  • Generative multi-modal Information Retrieval

  • Question Answering using First Order Logic

  • Neurosymbolic approaches

Research Interests:

  • Generative Question Answering

  • Information Retrieval

  • Graph Neural Networks

  • Neuro-symbolic systems

  • Differentiable end-to-end architectures

Publications

Retrieve-and-Rank End-to-End Summarization of Biomedical Studies

Gianluca Moro, Luca Ragazzi, Lorenzo Valgimigli, Lorenzo Molfetta

Similarity Search and Applications 2023