News
The latest news from our project team will be published in this section.
Participation in the Serbian-Chinese conference on the development of artificial intelligence
12.4.2024.
Yesterday, our team participated in a conference dedicated to the development of artificial intelligence, organized by the Science Fund of the Republic of Serbia and the National Science Foundation of the Republic of China. The aim of the conference was to introduce researchers from the two countries and to present previous projects in this field. We thank the Science Fund for being part of this interesting scientific event.
The second review meeting
3.4.2024.
š¢ The STOP project team met today, in the meeting room of the School of Electrical Engineering š« At the meeting, tasks for the upcoming quarter were divided š We are working diligently on the first hate speech detection software in the Serbian language, and we look forward to the new challenges ahead!
Formal ceremony of promotion of awarded projects from the project cycle "Program of young researchers - PROMIS 2023"
20.3.2024.
On Wednesday, March 20, 2024, a celebration was held on the occasion of five years of existence of the Science Fund of the Republic of Serbia. During the celebration, the attendees were addressed by the acting director of the Fund for Science Ms Milica ÄuriÄ-JoviÄiÄ, PhD, the Minister of Science, Technological Development, and Innovation, Ms Jelena BegoviÄ, PhD, as well as representatives of the European Union and the World Bank. At the celebration, managers of approved projects from the PROMIS 2023 call and projects that were awarded and received two-year funding from the Science Fund of the Republic of Serbia were especially highlighted.
Panel on cyber security trends and challenges
15.3.2024.
As part of the jubilee 30th national scientific and professional conference "YU INFO", our team organized a special event - an expert panel, on the topic "Cyber security trends and challenges in the protection of large systems", on Tuesday, March 12, 2024 in the congress hall "Pancic A" in the hotel "Grand" at the ski resort Kopaonik.
The panel was moderated and designed by the PI of our "STOP" project, Prof. Dr. Dražen DraÅ”koviÄ. The following researchers took part in this panel: prof. Dr. Slavko Gajin (Faculty of Mathematics in Belgrade), Ass. professor Dr. Maja VukasoviÄ (School of Electrical Engineering in Belgrade), Master of Legal Sciences Bojana MarinkoviÄ (University Newcastle, UK) and B.Sc. Eng. Marko Džida (Serbia And Montenegro Air Traffic Services Smatsa Llc).
In the past few years, cyber threats have become increasingly important for organizations and their large systems, and user data is very often under attack by malicious users (hackers). In the last year, many large software systems crashed in Serbia and the region. The panel discussed effective security measures, regulatory requirements (such as GDPR and HIPAA), good practices in organizations in Serbia and the world, the boundaries between the application of robust security measures and system maintenance, which data are dangerous to misuse, and many other questions. The panel attracted interested conference participants with interesting questions, so the discussion was fruitful and useful.
Participation of our researchers at the "YU INFO 2024" and "ICIST 2024" conferences
12.3.2024.
From March 10 to 13, 2024, the jubilee 30th ICT conference, "YU INFO 2024," and the 14th international conference, "ICIST 2024," were held at the ski resort Kopaonik, in which more than 350 participants from Serbia and the world took part. This year's conferences were mostly about artificial intelligence, big data processing, and large language models. At the conferences, 73 papers were presented in seven sessions at the national level and 92 papers at the international level. The conference program is available at the following link.
On Monday, March 10, 2024, in the scientific session "Pre-trained Large Language Models" at the international conference "ICIST 2024", our team presented a paper entitled "Enhancing Sentiment Analysis in Product Reviews: Fine-Tuning BERT for Class Imbalance and Optimal Sequence Representation", and on Tuesday, March 11, in the scientific session "Artificial Intelligence and Machine Learning" within the national conference "YU INFO 2024", a paper titled "AI-driven hate speech detection" was presented. Both scientific papers were created as a result of our project in the first two months of our research.
The kick-off meeting
10.1.2024.
In the meeting room of the School of Electrical Engineering, the STOP project team held an initial meeting on Wednesday, January 10, 2024. The Principal Investigator distributed the tasks to the members for the first month of the project and presented an overview of the half-yearly goals.
Project information
Acronym: STOP
The result of the researchers' collaboration will be a new software system that will detect hate speech in the Serbian language and will be of great importance in prevention of digital violence in Serbia.
Period: jan. 2024 - dec. 2025 | Budget: 140,000.00 €
The research problem addressed by this project is detecting hate speech (HS) in texts in Serbian on the Internet. Detecting and reducing HS is crucial for the safety and well-being of individuals, as it can otherwise lead to real-world harm and tragedies. Traditional methods of manually monitoring online content are time-consuming, costly, and ineffective in dealing with the vast amount of user-generated content. Therefore, there is a need for automated tools that can efficiently detect and prevent HS, which compose the primary goals of this project. This project also marks a significant milestone as the first-ever initiative to develop HS detection models designed exclusively for the Serbian language.
The impact of this project is remarkable. The software protects individuals from online abuse and violence by detecting and preventing HS. The project's research findings, including the developed dataset, NLP models, and software system, can advance the field of hate speech detection in the Serbian language and impact various sectors, including healthcare, education, science, and industry.
-
Participating Scientific-research organizations:
- University of Belgrade - School of Electrical Engineering (ETF)
Team members
The team consists of researchers from the Faculty of Electrical Engineering, University of Belgrade
Prof. Dražen DraÅ”koviÄ, PhD
Principal Investigator
Vladimir JocoviÄ, PhD
Member of the project team
Marko MiÄoviÄ, PhD candidate
Member of the project team
UroÅ” RadenkoviÄ, PhD candidate
Member of the project team
Jelica CincoviÄ, PhD candidate
Member of the project team
Adrian MilakoviÄ, PhD candidate
Member of the project teamResources
This section will show the resources that will be published during the project.
Published papers
Papers from conferences and scientific journals will be published in this section.
-
D.DraÅ”koviÄ et al., "Otkrivanje govora mržnje voÄeno veÅ”taÄkom inteligencijom"
D.DraÅ”koviÄ, V.JocoviÄ, A.MilakoviÄ, M.MiÄoviÄ, U.RadenkoviÄ, J.CincoviÄ, "Otkrivanje govora mržnje voÄeno veÅ”taÄkom inteligencijom", Zbornik radova 30. IKT konferencije "YU INFO 2024", Kopaonik, March 2024
Link: **Zbornik u produkciji!**
Apstrakt: āGovor mržnjeā predstavlja sve Å”to vreÄa pojedinca, populaciju ili pojavu na bilo kojoj osnovi, Å”to može da bude seksualna orijentacija, pripadnost nekoj religiji, nacionalnost, rasa, pripadnost odreÄenoj grupi, izreÄeni neki stavovi ili sliÄno. U ovom radu dat je pregled glavnih modela za detekciju govora mržnje koji su realizovani razliÄitim tehnikama veÅ”taÄke inteligencije, i prikazano je Å”ta Äe biti tematika projekta naÅ”eg istraživaÄkog tima. -
M.DodoviÄ et al., "Enhancing Sentiment Analysis in Product Reviews: Fine-Tuning BERT for Class Imbalance and Optimal Sequence Representation"
M.DodoviÄ, M.OgrizoviÄ, D.MiladinoviÄ, D.DraÅ”koviÄ, "Enhancing Sentiment Analysis in Product Reviews: Fine-Tuning BERT for Class Imbalance and Optimal Sequence Representation", Springer's Lecture Notes in Networks and Systems, with title Disruptive Information Technologies for a Smart Society (in publication), Kopaonik, March 2024
Link: **Proceedings in production!**
Apstrakt: Sentiment analysis, a pivotal aspect of NLP (Natural Language Processing), of-fers profound insights into the public sentiment from vast swathes of unstruc-tured textual data. This paper presents an empirical investigation into the applica-bility and effectiveness of the BERT (Bidirectional Encoder Representations from Transformers) algorithm for sentiment analysis, particularly focused on product reviews. The research delves into the nuances of consumer language expressions and evaluates the capacity of BERT to accurately classify sentiment in a large-scale dataset of food product reviews. The results achieved through this research are significant, with the fine-tuned BERT model demonstrating high accuracies, indicating its robustness and suitability for the sentiment classification task. In addressing the challenges posed by the varying lengths of consumer reviews, this study offers a methodological analysis for selecting the optimal max_seq_length parameter within BERTās framework. A critical balance is achieved between computational efficiency and the comprehensive inclusion of informative content within the reviews. Furthermore, the paper confronts the prevalent issue of class imbalance in sentiment analysis datasets by employing a weighted loss function during the training of BERT. This technique ensures equi-table representation and consideration of all sentiment classes, enhancing the model's accuracy and fairness.
Contact Us
Location:
Belgrade 11000, Bulevar kralja Aleksandra 73
Email:
stop@lists.etf.rs