AI · NLP · Trustworthy AI · Neuro-Symbolic AI · Digital Humanities

I study language, audio, and long‑context models.

I am Iacopo Ghinassi, an academic researcher working across natural language processing, multimodal learning, topic segmentation, and computational approaches to language and culture. I also teach, mentor students, and contribute to interdisciplinary digital humanities education.

View research Teaching and mentoring

About

Profile

My work focuses on making AI more accessible both for computer scientist practitioners and for non-technical individuals and humanities researchers. As such, I am currently working on neuro-symbolic approaches to trustworthy AI and I have previously worked on a range of topics in NLP from Multimodal representation learning to NLP methods applied to psychological and historical research.

Research direction

I am currently working on multimodal representation, uncertainty quantification and formal methods for deterministic verification of LLMs outputs.

I am especially interested in methods that remain useful outside ideal benchmark settings and in interdisciplinary applications that connect AI with cultural and social data.

Personal note

I play a variety of music instruments, all very poorly.
I used to be a decent tennis player.
I enjoy mentoring people on academic career and the AI topics I am expert on, outside of formal commitments (get in touch if you want to!)

Research

Selected publications

Work published in venues including COLING, LREC-COLING, EMNLP Findings, ACM ICMR, PeerJ Computer Science, and RANLP, with recurring themes around text segmentation, long-context LLM behaviour, and multimodal analysis.

Efficient Solutions For An Intriguing Failure of LLMs

COLING 2025 · Hosseini, Castro, Ghinassi & Purver. Long context windows do not automatically guarantee flawless analysis of long sequences. pp. 1880–1891.

Recent Trends in Linear Text Segmentation: A Survey

Findings of EMNLP 2024 · Ghinassi, Wang, Newell & Purver. A comprehensive survey of recent developments in linear text segmentation. pp. 3084–3095.

When Cohesion Lies in the Embedding Space

LREC-COLING 2024 · Ghinassi, Wang, Newell & Purver. Embedding-based reference-free metrics for topic segmentation. pp. 17525–17536.

MoralBERT: A Fine-Tuned Language Model for Capturing Moral Values in Social Discussions

GoodIT 2024 · Preniqi, Ghinassi, Ive, Saitis & Kalimeri. pp. 433–442.

Automatic Detection of Moral Values in Music Lyrics

ISMIR 2024 · Preniqi, Ghinassi, Ive, Kalimeri & Saitis. pp. 164–172.

Language Pivoting from Parallel Corpora for Word Sense Disambiguation of Historical Languages

LREC-COLING 2024 · Ghinassi, Tedeschi, Marongiu, Navigli & McGillivray. Case study on Latin. pp. 10073–10084.

Efficient Aspect-Based Summarization of Climate Change Reports with Small Language Models

NLP4PI @ EMNLP 2024 · Ghinassi, Catalano & Colella. pp. 123–139.

Multimodal Topic Segmentation of Podcast Shows with Pre-trained Neural Encoders

ACM ICMR 2023 · Ghinassi, Wang, Newell & Purver. pp. 602–606.

Lessons Learnt from Linear Text Segmentation

RANLP 2023 · Ghinassi, Wang, Newell & Purver. A fair comparison of architectural and sentence encoding strategies. pp. 408–418.

Comparing Neural Sentence Encoders for Topic Segmentation Across Domains

PeerJ Computer Science 2023 · Ghinassi, Wang, Newell & Purver. Not your typical text similarity task.

Exploring Pre-Trained Neural Audio Representations for Audio Topic Segmentation

IEEE ICME · Ghinassi, Purver, Phan & Newell.

Unsupervised Text Segmentation via Deep Sentence Encoders

DataTV @ ACM IMX 2021 · Ghinassi. A first step towards a common framework for text-based segmentation, summarization and indexing of media content.

Teaching

Instruction and mentorship

Alongside research, I have taught across core AI and data science subjects, delivered guest teaching in digital humanities, and supported students through mentoring and dissertation supervision.

Teaching assistant

Natural Language Processing

Teaching assistant · 2022, 2023

Natural Language Processing and Neural Networks

Teaching assistant · 2022, 2023, 2024

Machine Learning

Teaching assistant · 2022, 2023

Applied Statistics

Teaching assistant · 2024

Guest lecture & mentoring

Guest Lecture — King's College London

Introduction to Digital Humanities (postgraduate) · Digital Humanities MA

Delivered a lecture introducing what large language models are for Digital Humanities MA students.

Main Mentor & Cohort Point of Contact — QMUL

Data Science and Artificial Intelligence (Conversion Programme) MSc · 2022 cohort

Included mentoring support for MSc dissertation projects and pastoral and academic guidance for the full cohort.

Awards

Scholarship and recognition

Academic support and recognition have accompanied my teaching and research trajectory, including competitive scholarship-based enrichment opportunities.

Turing Enrichment Student Scholarship

The Alan Turing Institute · Recipient 2022/2023

Contact

Get in touch

If interested in knowing more about me, having a chat or ask me anything, find me at:

Email iacopo.ghinassi@ntu.edu.sg LinkedIn linkedin.com/in/iacopo-ghinassi-263280142 Google Scholar scholar.google.com — Iacopo Ghinassi