Carla Lopes and Sérgio Nunes (HumanISE)
“The EPISA project ended in December and Carla Teixeira Lopes (PI) and Sérgio Nunes (responsible for the development team) worked far beyond what’s required of members of an FCT research project.
This project brought together a team from INESC TEC, one from the University of Évora and one from the DGLAB (Arquivo Nacional da Torre do Tombo) – a total of 45 people throughout the project. The coordination of EPISA required great tenacity, since we went through a pandemic during a key stage of the project, with a significant rotation of grant holders (more than 20), aligning their contributions to the project with the requirements of academic degrees. Concerning the development of prototypes, the team dealt with technologies – ontologies, semantic networks, triplestores – with no mature development platforms, and designed a new generation of applications that address the multi-layered information of files, exploiting it in new ways”.
– HumanISE coordinators
What were the main challenges during the execution of the project?
We’d like to highlight two main challenges:
- The complexity and diversity of the technologies involved and the level of maturity of some of said technologies. We were forced to go the extra mile – more than we’d originally planned. This diversity and complexity at the technological level required a team available to explore and face many new problems.
- The difficulty in hiring Computer Science professionals led to a distribution of work by many grant holders. The dimension of the project and the number of grant holders involved over the years has been demanding, but also very rewarding.
How do you comment on the main results achieved?
We are very pleased with everything we have been able to achieve within the scope of the project.
In terms of practical results, we’d like to highlight the proposal and specification of ArchOnto, a linked data model for files, and the EPISA platform, a computational infrastructure that allows supporting applications based on linked data, designed and implemented in the project. This infrastructure allows to explore several use cases in terms of archives, namely the processing of linked data, the navigation and exploitation of said data, free text-based search, and access control. It’s worth emphasising relevant contributions to the interaction with linked data, namely through the publication of papers in major journals and conferences. Also important were the publication of data sets and code on repositories, the various academic works carried out, and the organisation of two international workshops. The results went far beyond what we initially anticipated and led to relevant opportunities for future work in these areas.
Which aspects of your job do you enjoy the most?
Being part of multidisciplinary teams to address real problems. The EPISA project was a good example, with a team that brought together Computer Science, Information Science, and Visual and Interaction Design. We’d also like to emphasise the contact with highly motivated and suited (actually, incredible) students and grant holders. Also important is dealing with new and complex problems, for which we develop and explore innovative technologies and solutions.
How do you comment on this nomination?
We are very pleased with this nomination because it means the recognition of the EPISA project and the entire team that participated in it. We’d also like to acknowledge the commitment and dedication to all EPISA grant holders.
Hugo Sousa (LIAAD)
“In March, Hugo published and presented a paper at the ACM SAC conference, which marked the conclusion of the MINE4HEALTH project, in partnership with IPO – very important to LIAAD and INESC TEC, which was also awarded at the Portugal Digital Awards. Despite joining this project while it was underway, Hugo brilliantly continued the work developed by Arian Pasquali (which was also nominated as an ‘Incredible’). Hugo was vital to the results of the project and the team’s cohesion. Moreover, Hugo carried out these activities while developing his PhD work, with tremendous dedication and joy. In addition, one of Hugo’s papers was accepted at SIGIR, a core A* conference, and he contributed to the organisation of the Tex2Story 2023 (April 1) as Web and Dissemination Chair. Hugo is a passionate and talented student, crucial to INESC TEC team as a researcher and colleague. He truly deserves this nomination – because he’s always incredible”.
– LIAAD coordinators
You presented a paper on ACM SAC conference, within the scope of the MINE4HEALTH project. How does it feel to participate in an awarded project, and which were the main results?
Being part of the MINE4HEALT project was an amazing and rewarding experience. Working with a talented team, dedicated to the development of solutions to improve the health and life of patients was extremely gratifying, particularly since the project addresses real world problems.
The main result of the project was the development of a pipeline that extracts medical entities (diseases, procedures, and medicine) from free texts written by IPO professionals. This pipeline in used to understand the medical record of patients, thus facilitating the decision-making process, and enabling the identification of relevant patterns.
Moreover, said identification has the potential to support the diagnosis and treatment of several diseases, contributing to a more personalised and efficient medical assistance to patients. This work demonstrates the power and relevance of combining medical knowledge and advanced natural language processing techniques to improve health care and people’s well-being.
Besides MINE4HEALTH, you’ve been working on your PhD. What is the focus of your work and what are the main conclusions?
My PhD project, “Unfolding the Temporal Structure of Narratives” focuses on exploring and improving the natural language models’ ability in terms of understanding the temporal aspect of narratives.
This research stems from the development of a framework to assess the models’ temporal abilities, named tieval (short for temporal information extraction evaluation). Said framework allowed us to observe that current models do not perform as expected in terms of temporal understanding. As to the upcoming phases, we’re analysing the importance of common sense in the comprehension of the temporal aspects of texts and exploring strategies to incorporate said knowledge into current models. This way, we aim to improve the natural language models’ ability to collect and interpret the temporal structure of narratives – which could have a significant impact on several applications, e.g., automatic generation of stories abstracts and collection of medical data.
Which aspects of your job do you enjoy the most?
I particularly enjoy the opportunity to work with talented and dedicated people, as well as the possibility to apply technical and scientific knowledge to address real world problems. I also like the constant learning and progression enabled by research activities, and the positive impact of our work on society.
How do you comment on this nomination?
This nomination is a recognition of my effort and dedication. I’m honoured to be nominated, and this is an extra motivation to keep working and contributing to the progress of science and technology in favour of society.
João Fernandes (CTM)
“The CTM coordinators would like to nominate researcher João Fernandes. João is a highly professional and motivated researcher, and he’s been developing a work with significant scientific impact, within the scope of his research activities. Although he’s currently initiating his PhD, he’s been demonstrating a great maturity, by taking on scientific activities with multiple CTM actions – including the participation in the organising teams of scientific events, the supervision of interns and master degree students, and the support to the execution of research projects. This nomination stems from João’s professionalism, availability and quality, namely during March, within the scope of the THEIA project – where he supported the work of younger colleagues and provided major scientific contributions, acknowledged by the project partners”.
– CTM coordinators
You’re currently initiating your PhD work. What is the focus of your work and what are the expected results?
My doctoral thesis plan focuses on applications in computational pathology and encompasses two main lines of research: weakly supervised learning and domain generalisation. Medical imaging data, in particular digital pathology, has a lot of variability and, in addition, clinical data are typically costly to acquire and annotate. Therefore, there is a gap between what fully supervised deep learning models demand, and what is the reality. Our proposal consists of researching and developing methods to learn representations that are less susceptible to domain-specific factors of variation (e.g., scanner, staining protocol, etc.) and that can learn with partial supervision. We aim to contribute towards the future implementation of computer assisted diagnosis (CAD) systems in daily practice of pathology labs.
You’re involved in the THEIA project; what kind of work have you’ve been developing and which are the main challenges?
Within the THEIA project, I focused on researching more energy-efficient algorithms, namely Spiking Neural Networks (SNNs). These algorithms have advantages over conventional deep learning models since their activation functions simulate biological neurons, i.e., binary activation functions, and that allow asynchronous processing, thus being ideal for event-based applications, such as autonomous driving. However, there is a gap between what is understood theoretically and practical applications, which raises some technical difficulties. I would say that the main challenges are SNNs being limited to shallower and less complex models, the need to carefully tune all hyperparameters, and the limited availability of neuromorphic hardware accelerators (the equivalent of GPUs in conventional deep learning) that would allow assessing the real energy efficiency of SNNs when compared with conventional deep learning.
Which aspects of your job do you appreciate the most?
I would like to highlight two main points: on the one hand, the constant learning, characteristic of the scientific community. On the other hand, the freedom to get creative, both in the formulation of scientific hypotheses and the presentation and discussion of results.
How do you comment on this nomination?
I interpret it as an incentive to keep following many of the work practices that I have been adopting so far. At the same time, I would like to mention the contribution of my advisors, group colleagues and, of course, the VCMI and CTM coordinators. The good work environment, the collaboration and dialogues, as well as the different critical perspectives, contribute to both the individual success and to the scientific success of the group/research centre. These people are often a source of inspiration.
Miguel Correia (HumanISE)
“The HumanISE coordinators would like to nominate Miguel Correia. Over the last few months, Miguel has demonstrated an exceptional dedication and tremendous team spirit, which were relevant to the high technical-scientific quality of the results of the MELOA, COL4INDLOG and iReceptor+ projects – and are still vital to the ongoing ILIAD project. It’s important to mention his direct contribution to the definition, development, and documentation of results of the MELOA project. Characterised by volatile requirements, short development cycles and a small team, the project led to five Pre-Disclosure Forms, and the results are being successfully used in campaign management to obtain in-situ ocean data in (near) real-time”.
– HumanISE coordinators
You were involved in the MELOA project, particularly in the definition, development, and documentation of results. What were the main results of this project?
During the MELOA project, we had to overcome several challenges, leading to a wide diversity of applications and functionalities. One of the main results was the WAVY Operation Software (WOS) platform, which allows the management of campaigns, equipment, and datasets. In addition to WOS, we developed another platform for Citizen and Science, which allows anyone to track floats in public campaigns, in real time. Another important result was the creation of a mobile application to support the collection of equipment, since the floats are small and often difficult to find in the open sea. Overall, the MELOA project was extremely enriching and enabled the development of important technologies to monitor and manage campaigns in marine research operations.
Some of the results are being used in campaign management to obtain in-situ ocean data. How do you perceive the application of project’s results in real contexts?
It is always gratifying to see that the results obtained during the MELOA project are used in campaign management. This is a sign that the work we do has not been restricted to a controlled research environment but has a real and positive impact on the world. When project results are used in real contexts, this validates the research work, with a significant impact on different areas. Concerning this project, the results may be useful in future studies or other projects dedicated to the ocean, thus contributing to the advancement of science and technology.
Which aspects of your job do you enjoy the most?
One of the things I appreciate about my work is the diversity of projects and tasks, which constantly challenge me. This variety prevents the work from becoming monotonous and promotes continuous learning – keeping me updated on technologies and tools.
How do you comment on this nomination?
Although I was surprised, I am proud to know that my work and dedication are recognised.
Paula Rodrigues (HASLab)
“Paula Rodrigues’ work in the dissemination and promotion of HASLab activities is always worthy of recognition, due to her motivation and professionalism. In recent times, and in addition to these tasks, Paula contributed significantly to the coordination and participation of INESC TEC in two European project proposals: MARIBEL and HANAMI – while contributing to the communication plans of said projects. The MARIBEL project is a CSA between Europe, Latin American and Caribbean countries that aims to leverage the objectives of the BELLA network, while consolidating established R&D cooperation networks, promoting EU values in people-centred digital transformation, and contributing to sustainable development goals. The HANAMI project is a collaboration between major European and Japanese supercomputing centres; it focuses on preparing said facilities for the challenges of the new generation of supercomputers – the exascale”.
– HASLab coordinators
You’re involved in the coordination and participation of INESC TEC in two European project proposals, contributing to their communication plans. What are the main challenges related to communicating European projects?
The communication and dissemination activities of project results are increasingly crucial throughout the course of the project; hence, it is good to know that these elements are already an important requirement when designing a proposal. At this stage, the challenges are to understand the objectives of the project, analyse the target-audiences and resort to our creativity in the planning of the various communication actions over the years of execution. To me, the best part is to think about the various communication actions to be carried out, always resorting to digital marketing, advertising, public relations, and even direct marketing tools. All projects are different, and this “tailor-made” approach is a real challenge!
What motivates you the most in the dissemination of the work developed within the scope of these projects?
Based on my experience, which essentially focuses on projects in Computer Science, I often feel admiration and recognition by the international partners throughout the course of the various communication actions – probably because they are technical partners, with proven expertise in Computer Science. Moreover, I appreciate the regular contact with different communities and discussing ideas stemming from these relationships.
Which aspects of your job do you enjoy the most?
I joined INESC TEC in 2014, and I must say that every day was – and still is – completely different. Every week, there is something new to do and that is quite motivating and challenging.
How do you comment on this nomination?
The proposals required significant investment but will certainly be rewarded when they are approved. This nomination is, by itself, a recognition of said effort and commitment; it is always good to know that I have done a positive job for the Institute.
Pedro Amorim (CEGI)
“This nomination stems from the publication of two research papers in FT50 journals, which is a remarkable achievement. The first paper, published in Manufacturing Services & Operations Management (MSOM), presents a comprehensive analysis of the impact of delivery consolidation on consumer behaviour, in the context of online retail. This paper contributes significantly to the supply chain management literature and highlights the importance of logistics and delivery strategies to improve customer satisfaction and reduce returns. The publication in a high-profile journal like MSOM proves the quality of research and the relevance of results – for both academics and professionals. The second paper, published in MIT Sloan Management Review, is a translation of the first work for managers, making it accessible to a wider audience. The MIT Sloan Management Review focuses on bridging the gap between theory and practice in the field of management. The work in this journal demonstrates the authors’ ability to effectively communicate research to both academic and business audiences, which is a valuable skill within the scope of INESC TEC. Globally, the publication of these two papers in FT50 journals reinforces CEGI’s position”.
– CEGI coordinators
You’ve recently published two research papers in FT50 magazines. The first contributes to the supply chain management literature and highlights the importance of logistics and delivery strategies to improve customer satisfaction and reduce returns. Can you tell us more about the results of your analysis and the main research conclusions?
We noticed that delivering all products in a single order, even if later, reduces the likelihood of a return, which improves the financial performance of companies and their suppliers and reduces reverse logistics. Our results suggest that, in this context, speed of delivery is less important than the convenience of receiving all ordered items in a single delivery, and we provide instructions for adapting logistics strategies accordingly.
The second paper, published in MIT Sloan Management Review, is a translation of the first work for managers, making it accessible to a broader audience. What were the main challenges you faced with this paper?
I would say we faced two main challenges: shortening all reasoning to a 2500-word executive format and convincing the editor that our results could be of interest to a wider audience.
Which aspects of your job do you enjoy the most?
The novelty factor and multidisciplinarity of our results. Our findings are counterintuitive to most retail managers, who typically prioritise fast delivery based on the assumption that what customers want most are quick rewards. But there is one psychological phenomenon that may explain why parcel deliveries can cause more returns: the sense of incompleteness caused by a disaggregated order. We showed that this effect is triggered by the Zeigarnik effect, named after psychologist Bluma Zeigarnik, who observed that our brains constantly remind us of unfinished tasks, while completed tasks are quickly forgotten.
How do you comment on this nomination?
I was very grateful for this acknowledgment. Publishing in journals of great impact and dissemination is one of the objectives of our Centre (and of INESC TEC), and it’s good to be able to make a small contribution in this regard.
News, current topics, curiosities and so much more about INESC TEC and its community!
