Researcher and professor at the University of Évora, Paulo Quaresma, shares potential applications of artificial intelligence technologies in FCCN Unit services.
This article aims to reflect on how the application of artificial intelligence methodologies can enhance and increase the services provided to the scientific community by FCCN, a unit of FCT - Foundation for Science and Technology.
These reflections are based, on the one hand, on my knowledge of FCCN, as I have been a member of FCT's Board of Directors during the year 2021, and, on the other hand, on my 25 years of research in the field of artificial intelligence.
Let us begin by characterizing FCCN's areas of activity. To this end, a consultation of its website(http://www.fccn.pt) allows us to identify a wide range of services provided to the scientific community. While it is true that some of these services already incorporate components of artificial intelligence, it is expected that, in the short and medium term, they may be extended by incorporating components of high added value. Let's look at some possibilities.
The historical archive of the .pt domain web pages is, by itself, a collection of immense value and a base for scientific research studies, especially in the area of computational processing of Natural Language and Computational Linguistics. However, with the increasing availability of open access tools for processing Portuguese language texts, the "Arquivo.pt" may make available not only the archive, but also services over this archive. As an example, see the potential interest in automatically obtaining the identification of the entities (people, organizations and places) mentioned in the publications of certain domains (e.g. online newspapers) in a given period of time; or, in addition, which events the people concerned participated in and with whom they related. Or, in what ways has the lexicon and syntax used in online publications changed over time?
- B-on: Knowledge Library
b-on integrates a vast set of scientific publications from the main international publishers and allows an integrated search to those collections. However, it is possible to develop and integrate advanced Natural Language processing tools, in order to incorporate a set of additional functionalities in the search module in the article collections. In fact, Language models are available nowadays, both for English and multi-languages, which may be used in the searches, allowing to obtain the scientific articles that are semantically closest to the searches performed. It is also possible to build a model of each b-on user, with their scientific interests, and send notices whenever an article related to their interests is added to one of b-on's integral collections.
- Open Science
Portugal joined plan S in 2021, with the goal of adopting the principles of Open Science, either in the publication of scientific articles or in the creation of data repositories. In this sense, FCT has been incorporating in its calls for proposals measures for the full adoption of plan S during this year. However, the monitoring and evaluation of the procedures associated with Open Science is not simple, given the large size of the data at stake (annual production of thousands of articles and data collections). The use of AI and Data Science methodologies may allow to automatically extract the relevant data and validate their publication form, eliminating the need for complex and repetitive manual procedures.
- Advanced Computing
This is the item where the relevance of artificial intelligence is most immediately obvious for its implementation and enhancement: machine learning and, specifically,deeplearning require increasingly powerful and complex computational resources. FCCN/FCT manages the RNCA - National Network for Advanced Computing, which consists of several nodes at national level and includes structures dedicated to high performance processing, including deep neural networks. A huge increase in the existing computational capacity is expected in the short and medium term.
- Researcher Services
FCCN provides a wide range of services to researchers, from services for content production (educast), teaching and training platforms (NAU), video conferencing systems (colibri) and broadcasting of events (videocast), to platforms to support the management of scientific activity (ciência id, ciência vitae and PTCRIS). Also in this area, artificial intelligence may help to raise the level of services available to the scientific community. Some examples: the teaching and training platform can benefit from the incorporation of trainees' models, with the representation of their mental state (e.g. goals and beliefs), in order to allow an interaction and a personalized learning process; the CV management platform could, through Natural Language processing tools, identify additional relevant information in external sources and propose its inclusion in the corresponding CVs.
- Infrastructures: RCTS, networks
This is a fundamental component of FCCN's activity and supports the activity of the entire scientific community. It is an area that already integrates AI components, such as intrusion and attack detection or anomaly diagnosis. With the growing importance of cybersecurity, AI will have an increasingly relevant role in this domain.
In conclusion, from the analysis carried out, as well as thanks to current developments in the area of AI, it is clear that this will be an inevitable and successful intersection, allowing FCCN to increase the quality of the services it already provides to the scientific community.