Hugo Sousa

PhD Student of Machine Learning @ University of Porto

Machine Learning Researcher @ INESC TEC

Teaching Assistant @ University of Porto

I am a PhD student at the University of Porto. My research interest is mainly focused on machine learning with a specific focus on natural language processing. In particular, the focus of my current research - and of my PhD - is on the temporal reasoning capabilities of language models. That is, how do language models "understand" and "manipulate" temporal information (I try to be careful with words). Besides that, I am a deep reinforcement learning `aficionado` despite all its challenging and complex features.

I am also a research assistant at LIAAD, the AI lab from INESC TEC, where I am frequently associated in some project to build NLP solutions to some practical problems. As my last ongoing work, I am a teaching assistant of the Machine Learning course at the University of Porto.

My bachelor's is in Physics and my master's is in Applied Mathematics, so we can talk about that if that is of your interest.

Work

Visting PhD Student: @ Carnegie Mellon University
October 2023 to December 2023
Teaching Assistant: @ University Porto
February 2023 to …
PhD Candidate: @ University Porto
December 2020 to …
Research Assistant: @ INESC TEC
December 2020 to …
Data Scientist: @ BNP Paribas
December 2019 to December 2020
Data Scientist: @ JTA: The Data Scientists
July 2018 to April 2019

Education

2020-2024 (expected): PhD, Computer Science; University of Porto
As an FCT Grant holder and part of the Text2Story project. Advised by Professor Alípio Jorge and Professor Ricardo Campos.
2017-2019: MS, Applied Mathematics; University of Porto
Thesis title: ECG Compression and QRS Detection: an IoT Approach
2014-2017: BSc, Physics; University of Porto

Publications

Sousa, H., Guimarães, N., Jorge, A., and Campos, R. GPT Struct Me: Probing GPT Models on Narrative Entity Extraction WI-IAT, 26-29 October 2023, Venice, Italy

Sousa, H., Campos, R., and Jorge, A. TEI2GO: A Multilingual Approach for Fast Temporal Expression Identification CIKM, 21-25 October 2023, Birmingham, United Kingdom

Sousa, H., Jorge, A., and Campos, R. tieval: An Evaluation Framework for Temporal Information Extraction Systems SIGIR, 23-27 July 2023, Taipei, Taiwan

Sousa, H., Jorge, A., Pasquali, A., Santos, C., and Lopes, M. A Biomedical Entity Extraction Pipeline for Oncology Health Records in Portuguese ACM SAC, 27-31 March 2023, Tallinn, Estonia

Sousa, H. Temporal Relation Extraction: The Event Ordering Task DESIRES, 15-18 September 2021, Padua, Italy

Sousa, H. ECG Compression and QRS Detection: an IoT Approach Master Thesis, 2019

Other

🥉 Sword AI Challenge 15-23 July 2023
Member of the team - Team Physio - that took 3º place on Sword AI Challenge 2023.

IACT’23 @SIGIR 27 July 2023
Web and Dissemination Chair of the IACT Workshop.

Text2Story'23 @ECIR 2 April 2023
Web and Dissemination Chair of the Text2Story Workshop.

ESSIR 2022 18-22 July 2022
Attended the European Summer School in Information Retrieval.

Text2Story'22 @ECIR 10 April 2022
Web and Dissemination Chair of the Text2Story Workshop.

🥇 GENTIL Project 2021
Part of the team that developed the natural language processing pipeline for the GENTIL porject. This project was recognized as the "Best Future of Work Project" on Portugal Digital Awards 2021.

DSAA21 6-9 October 2021
Volunteer.

DESIRES 2021 15-18 September 2021
Went to the beautiful city of Padua to participate on the DESIRES 2021 conference.

LxMLS 2021 7-15 July 2021
Attended the Lisbon Summer School 2021.

🥈 Eurekathon 2020 5-7 November 2020
Member of the team - Feeding the Future - that took 2º place on Eurekathon 2020.