%T  International Conference on User-System Interaction - ICUSI 2024
%C  Constanta
%D  19-20 September 2024
%G  ISSN 3061-3205
%E  Crenguţa-Mădălina Puchianu
%E  Dragoş Datcu
%P  215
%W  https://rochi.utcluj.ro/icusi2024

%T Explainable Artificial Intelligence Techniques for Extended Reality Systems: a Systematic Literature Review
%A Clara Maathuis
%A Marina Anca Cidota
%A Dragoș Datcu
%A Letiția Marin
%P 2-9
%X The integration of Artificial Intelligence (AI) in the Extended Reality (XR) enables the development of more immersive, interactive, and personalized environments and significantly enhances user experiences. Further, ensuring transparency of data, reasoning, and decision-making processes underlying the AI technologies used, represent critical aspects that require the development and deployment of relevant Explainable AI (XAI) techniques. Implementing such mechanisms not only ensures compliance with existing social and ethical norms and values, but also fosters user trust and supports the further adoption of such solutions. While both research and practitioner efforts are seen in this direction, there is a gap in the existing body of knowledge regarding a review of existing XAI methods and techniques tailored to XR systems. To address this point, this paper presents a systematic literature review of XAI approaches applied in the XR domain to synthesize current research, identify existing trends, and highlight potential future research areas.
%K explainable AI; responsible AI; trustworthy AI; Extended Reality; Mixed Reality; Augmented Reality; Virtual Reality
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p002-009.pdf

%T A P300-based Brain-Computer Interface to Control a 3D LabVIEW Simulation using GTEC Unicorn P300 Speller Aimed at Cognitive Training
%A Oana Andreea Rușanu
%P 10-13
%X The Brain-Computer Interface (BCI) is an innovative, multidisciplinary research field that offers cutting-edge technologies in medical engineering. BCIs assist individuals with neuromotor disabilities in controlling assistive devices or enable mental communication through the P300 Speller interface. Additionally, BCIs provide opportunities for cognitive training for individuals suffering from cerebral strokes or spinal cord injuries by controlling interactive virtual simulations. This paper presents a novel approach for integrating the LabVIEW graphical environment with the GTEC P300 Unicorn Speller to control a 3D scooter guided by a humanoid robot across various movement directions. The proposed brain-computer interface utilizes the Unicorn headset equipped with 8 EEG sensors. Three subjects participated in experiments involving 16 movement commands across three difficulty levels.
%K Brain-Computer Interface; P300 Evoked Biopotential; LabVIEW Instrument; Unicorn EEG headset.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p010-013.pdf

%T Generative AI System to Support Decision-Making on Public Procurement Legislation by Fine-Tuned Large Language Models
%A Dragoș Datcu
%A Clara Maathuis
%P 14-21
%X Public procurement processes are often hindered by the complex nature of implementing domain-specific directives and laws, leading to human errors in decision-making during acquisition tasks. These errors can significantly impact the timely execution of national and European government projects. Given recent developments in the Generative AI (Artificial Intelligence) domain, to address this challenge, we present an innovative LLMs-based (Large Language Models) system containing a teacher-student architecture based on LLaMA 3- 70B as teacher model and a series of student models like RoMistral-7B, LLaMA 3-8B, Mistral- 7B, Gemma-7B and Saul-7B specifically adapted for interpreting and reasoning on Romanian public procurement legislation.

This explorative solution is aimed to support legal experts in navigating the intricacies of procurement laws and directives, thereby enhancing decision-making accuracy and reducing errors in public acquisition tasks. The solution is developed based on brainstorming sessions conducted with experts in Romanian public procurement. To this end, a small-scale experiment is conducted to evaluate the approach by employing both quantitative and qualitative analysis methods. The results of our preliminary study demonstrate remarkable performance in understanding and reasoning on the public procurement corpus, with the fine-tuned model showing promising capabilities in interpreting complex legal texts and providing valuable insights for procurement professionals.
%K LLM; Llama3 fine-tuning; Legal corpora.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p014-021.pdf

%T Travel Me Now - Convolutional Neural Networks in Augmented Reality Exploration of Paris
%A Miruna-Elena Iosub
%A Adrian Iftene
%P 22-29
%X Augmented reality (AR) is being increasingly utilized across many diverse fields. Its applications span medicine, education, gaming, interior and exterior design, attracting a wide range of age groups, particularly young people excited by technological innovations. Today, several AR applications are available to help tourists navigate cities and uncover interesting information interactively. In this paper, we propose an application that can recognize certain tourist attractions from Paris, France more exactly- the Eiffel Tower, the Louvre Pyramids, the Arc de Triomphe, the Notre-Dame de Paris, and the Orsay Museum with the help of AR and with the help of convolutional neural networks (CNNs). Using AR, once the user has arrived at the tourist destination and classified it with the help of CNN, they will have an immersive experience in which they will learn various curiosities and historical details about the place and the visiting schedule.
%K Augmented reality; convolutional neural networks; mobile application
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p022-029.pdf

%T Recognition of Romanian Sign Language Alphabet Using Convolutional Neural Networks
%A Emilia-Maria Nuță
%A Gheorghe Cosmin Silaghi
%P 30-33
%X The recognition of sign languages is critical for facilitating communication between the deaf and hearing communities. This paper addresses the problem of recognizing the Romanian sign language (RSL) alphabet using convolutional neural networks. We trained both custom made and transfer learning models. Due to non-availability of datasets on RSL, we recorded a dataset from two different subjects performing 26 signs in 4 different lightnings. Comparative analysis demonstrates that leveraging transfer learning outperforms the custom models in terms of accuracy. Key contribution of this paper is the development of a comprehensive dataset for the RSL alphabet that was not available before, facilitating the creation of other models in this area. Our findings underscore the potential of deep learning in advancing assistive technologies.
%K Romanian sign language; Convolutional neural networks; Image classification; Deep learning
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p030-033.pdf

%T SmartReviewHub: An AI Based Scientific Conference Management System
%A Mihail Popescu
%A Paul-Stefan Popescu
%A Marian Cristian Mihaescu
%P 34-40
%X Conference management systems are one of the most essential tools for evaluating papers submitted to conferences or journals. These systems evolved over the years, providing better functionalities to facilitate the conference management process. The problem tackled by this paper is to provide a system that enhances the user experience and ease of use for all the actors involved. The approach was to integrate machine learning and deep learning features in order to provide features like concept extraction, plagiarism detection, spell checker or even sentiment analysis. The result is an intelligent conference system that can provide all the classic and new functionalities for an optimised user experience.
%K Conference system, machine learning, deep learning, user experience.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p034-040.pdf

%T Indoor Navigation System Using Augmented Reality Technology at the Technical University of Cluj-Napoca
%A Dragoș Andrei Andreșan
%A Călin Gheorghe Dan Neamțu
%A Radu Comes
%P 41-48
%X This paper presents an augmented reality application designed to enhance indoor navigation at the Technical University of Cluj-Napoca. The case study focuses on an indoor navigation system to assist students and visitors in navigating from one room to another within the 4th floor of the "M" building. Given the numerous classrooms in the faculty building on Muncii Boulevard, navigating without prior knowledge of the layout was time-consuming. A 3D model of the 4th floor was created using Autodesk Revit and exported into Unity to integrate seamlessly into an augmented reality (AR) application. The AR application was developed using Unity along with the "AT+Sync | Synchronize with Unity," "AT+Sync v2 | Synchronize from Revit," and "AT+Explore | Interactive Archviz" plugins. This application allows students and visitors to navigate more efficiently within the Technical University of Cluj- Napoca and can be easily adapted for use in other buildings. By leveraging the versatility of these tools, the application not only improves way finding but also has the potential to enhance overall user experience, reduce stress associated with finding rooms, and increase the accessibility of university facilities for newcomers and guests.
%K Augmented reality; unity; indoor navigation; computer- aided design;
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p041-048.pdf

%T Using Virtual Reality in Promoting Cultural Heritage
%A Adrian Topoleanu
%A Dorin-Mircea Popovici
%P 49-56
%X This paper details the creation and implementation of a virtual reality (VR) museum aimed at promoting cultural values and heritage, specifically focusing on the Dobrogea region in the southeast of Romania. Based on Unreal Engine 5 [38] and VR technologies, Dobrogea Virtual Reality Museum offers an interesting, interactive learning experience that departs from traditional museum visits. The region, divided into Tulcea and Constanța counties, is an underappreciated hub of art and culture. The VR museum addresses a significant issue faced by local museums: lack of public interest and the need to adapt to digital advancements. By providing an immersive, accessible platform, the application aims to revitalize these cultural institutions. Users can explore photorealistic representations of real artifacts, captured using photogrammetry techniques, enhancing visual fidelity. Informative text displays taken directly from museum staff complement the experience, making learning a more engaging experience. Additionally, the VR museum allows users to interact with and manipulate cultural objects, creating a personal and memorable encounter.
%K Human-centered computing: Human computer interaction (HCI); Interaction design; Visualization; Cultural Heritage accessibility;
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p049-056.pdf

%T StressVR: Visual Analysis Tool for ECG and HR Data in Virtual Reality
%A Raul-Alexandru Gorgan
%A Adrian Sabou
%A Teodor-Traian Ștefănuț
%P 57-64
%X Analyzing large datasets from physiological sensors, such as ECG (Electrocardiogram) and HR (Heart Rate), presents significant challenges with traditional 2D visualization methods. These methods often fail to reveal detailed patterns and connections within the data, making it difficult for researchers to draw meaningful conclusions. StressVR addresses these challenges by offering a visual analysis tool that use Virtual Reality to create an immersive 3D environment for data exploration. With StressVR, users can interact with data in a more intuitive and engaging manner, enabling the selection and examination of individual data points, intervals, and multiple datasets simultaneously. This tool also offers advanced features such as data filtering, removal and annotation, improving the ability of the user to detect patterns and anomalies. Navigation between data points is possible, offering a clear advantage over the 2D approach. The evaluation of StressVR demonstrates that the immersive 3D approach improves the accuracy of data analysis compared to traditional 2D methods. The results indicate that StressVR not only improves the understanding of physiological responses to stress but also provides a more efficient and user-friendly platform for analyzing large-scale sensor data. This paper details the development, experiments and evaluation of StressVR, highlighting its importance in the field of physiological data analysis.
%K HCI; Virtual Reality; Visual Analytics; Stress recognition; Heart rate variability; ECG; Unity;
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p057-064.pdf

%T HCI: Research into the Effects XR has on Users, an Exploratory Study on Apple Vision Pro
%A Panagiotis-Efstratios Chontas
%A Adrian Iftene
%A Sabin-Corneliu Buraga
%P 65-71
%X Recent increases in Extended Reality (XR) headset sales suggest a growing consumer preference for Augmented Reality (AR)/Virtual Reality (VR) in home and institutional environments. As XR technology becomes mainstream and evolves, it is essential to examine its physical and mental impacts. While recent studies emphasize the positive effects of XR, this paper argues for investigating potential negative outcomes to ensure user well-being and ethical standards. This study involved young adults (ages 18-29) using the Apple Vision Pro headset within a visionOS environment, designed to encompass a range of user interactions. Questionnaires and health tests were conducted to evaluate participants. Findings indicate users’ adeptness in interacting with XR technologies, particularly favoring experiences that simulate real-life environments in three dimensions. However, despite their familiarity with technology and extensive screen time, young adults experienced physical side effects early in the testing process. This research underscores the necessity of a balanced understanding of XR’s implications to guide its development responsibly.
%K Human Computer Interaction (HCI); Extended Reality (XR); visionOS; Apple Vision Pro; Augmented Reality (AR); Virtual Reality (VR); User Interface (UI).
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p065-071.pdf

%T Design and Implementation of a 2D Game Engine: Algorithmic Approaches and Performance Optimization
%A Maria Pasca
%A Constantin Nandra
%P 72-79
%X This paper describes a solution for generating pseudo-random mazes to act as environments for a puzzle-type game that requires the player to make decisions while being chased by one or more opponents. We focused on developing a flexible and scalable solution, aiming to minimize the time required for the generation process. Throughout the paper, we will describe the inner workings of the proposed solution, while focusing on the optimization of the maze generation and path finding aspects. We also approach the problem from the perspective of path finding, which is relevant for the implementation of the intelligent agent chasing the player. Finally, we present the findings of a set of tests done on the described solution, which was utilized in the implementation of a maze-chase game, featuring NPC navigation through the generated maze. These include both user and system performance tests, as well as a user satisfaction survey.
%K maze generation algorithms; dynamic path computing; game development
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p072-079.pdf

%T Simfonia Silentio: The Impact of the Exclusively Auditory Game in Creating Equality and Innovation in the Gaming Universe
%A Patrick Antony Tudorie
%A Dorin-Mircea Popovici
%P 80-86
%X Our paper focuses on the development and influence of a game, named “Simfonia Silentio”, meant for visually impaired people, and illustrates the ways in which auditory games can be used as tools for promoting equality and innovation within the video game industry through increased accessibility and a new way of experiencing games. The current context in which digital accessibility and inclusion are becoming increasingly relevant is what makes this project important. “Simfonia Silentio” goes beyond by being a novel approach to assisting people with visual impairment; it also demonstrates how technology can be used for providing fair and innovative gaming experiences. Moreover, we proved the positive impact of auditory games on visually impaired users through the evaluation and testing of an accessible game, suggesting a new direction in designing video games. By bringing out both problems and solutions faced during project development, we give suggestions on similar future projects and stress out the importance of incorporating accessibility when designing video games. This way, “Simfonia Silentio” represents an essential step towards accessible games that embrace equality and inventiveness.
%K Auditory games; accessibility; sound design; user experience; web technologies; game engine; immersive experience; inclusive gaming; innovation in games.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p080-086.pdf

%T Spatial Big Data Visualization and Manipulation
%A Maria Gliga-Hambet
%A Dorian Gorgan
%P 87-94
%X This paper explores the visualization and manipulation of spatial big data in a Virtual Reality (VR) environment. The solution allows users to interactively visualize, manipulate and understand complex datasets. It aims to improve interpretability of data by leveraging the immersive qualities of VR. The tasks involve describing sequences of operators that work together to achieve a specific user goal. Experimental validation evaluates these operators in terms of functionality, usability, performance, and scalability.
%K Multidimensional data; data visualization; virtual reality (VR); data manipulation; user experience (UX).
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p087-094.pdf

%T Computer-support for the analysis of rhetorical devices in texts
%A Ștefan Trăușan-Matu
%A Adrian-Dinu Urse
%P 95-98
%X In this paper we present the background ideas and theories of an implemented prototype of an artificial intelligence, natural language processing application that takes a text as input and outputs the analysis of several classes of repetitions, which are types of rhetorical devices. The application is targeted for students, for researchers, and for any person that aims to improve writing or oratorically skills. It was used for analysing chapters from two well-known novels and speeches, the paper containing also excerpts of some of the results.
%K Artificial intelligence; natural language processing; text analysis; rhetoric devices; phonaesthetics
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p095-098.pdf

%T Enhancing Natural Human Interactions in AR/VR Applications
%A George-Gabriel Constantinescu
%A Adrian Iftene
%P 99-106
%X In the evolving landscape of immersive technologies, enhancing natural human interactions within AR/VR applications have a crucial role for creating intuitive and engaging user experiences. This article focuses on various innovative interaction techniques, including teleportation, tool manipulation, magic movement, and long grabbing functionality. By leveraging advanced VR controllers, hand recognition, and user-focused hand movements, we aim to bridge the gap between the virtual and physical worlds, promoting more natural and seamless interactions. Our research examines different user interface types and interaction modalities, showcasing how these methods can significantly enhance user engagement, improve task performance, and provide a more immersive and realistic experience in virtual environments. This study highlights the transformative potential of these technologies in the field of user-system interaction, setting the stage for future advancements in AR/VR applications.
%K Virtual Reality (VR); Augmented Reality (AR); User Experience (UX); User Interaction; Human-Computer Interaction (HCI)
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p099-106.pdf

%T Rhythm analysis of Romanian texts
%A Onur-Șener Veis
%A Ștefan Trăușan-Matu
%P 107-113
%X The paper presents an implemented interactive application for the analysis of the rhythmicity of texts in the Romanian language, using methods based on artificial intelligence. For the text analysis, are used three mathematical methods developed by Solomon Marcus, Mihai Dinu, and Vasile Vasile. The application was used for analyzing four genres of text: poems, prose, speech, and scientific text.
%K Rhythm; accent; syllable; natural language processing; artificial intelligence
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p107-113.pdf

%T Interactive Physics Experiments using Gesture Recognition
%A Mario-Cristian Constantinescu
%A Anata-Flavia Ionescu
%P 114-121
%X In this paper we propose a web application which provides virtual and interactive physics experiments for secondary school students. The application allows users to perform actions such as creating and manipulating objects in the virtual experiments using both mouse/keyboard and hand gesture input. Within each experiment, supported actions are associated with default gestures, but each user is allowed to customize his/her interaction with the experiment by permuting the associated gestures.

Gesture recognition is performed using a deep neural network classifier with 18 classes of gestures. The classifier is pretrained on a custom dataset of 368 samples. Each sample is a set of 21 hand landmarks extracted using Google’s MediaPipe Hands framework. Users are also allowed to fine-tune the weights of the model by adding samples of their own gestures for the supported classes.

The application has the potential of enhancing the quality of experience of students, offering customizable multimodal interaction and the possibility to create affordances. We conclude the paper with a discussion of the contributions, limitations and future research directions.
%K virtual learning; gesture recognition; neural network; interactive simulation; physics; motivation; self-paced learning; MediaPipe
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p114-121.pdf

%T Analysis of the structure similarity of musical scores
%A Andrei-Daniel Ghițeanu
%A Ștefan Trăușan-Matu
%P 122-129
%X The field of music analysis has expanded significantly in response to recent developments in artificial intelligence and machine learning. The research presented herein is focused on the study of resemblance between musical scores, therefore exploring different techniques used in calculating similarity ratings. The processing starts by representing the musical data extracted from song files in a manner that facilitates analysis and presents advantages in depicting the relationships between the elements present in the compositional structure of the musical pieces, leading to the use of a knowledge graph for this purpose. Three proposed methods will be inspected for calculating the ratings of comparison between different melodies and music measures, based on generating embeddings, music criteria and integration. The aim is to verify the reliability for integration of those approaches while developing more complex applications.
%K Musical score analysis, ontology, music similarity, artificial intelligence.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p122-129.pdf

%T S.T.A.R.S. - Stellar Teaching and Astronomical Realism Simulator
%A Octavian-Mihai Matei
%A Adrian Sabou
%P 130-136
%X This paper describes the development of a physicsbased simulation application for educa- tional purposes. We propose a solution that has the main aim to enhance the visual learning experience of students during Physics classes through the simulation of the behaviors of stars using Newtonian-based formulae in different scenarios. The target users will be high school students enrolled in the 9-12th grade in the physics classes. The users will be able to get real-time information about different statis- tics of real-world stars using the ESA GAIA database. This research paper presents the de- velopment steps, beginning with the conceptu- alization of the application’s scope and con- text, it progresses through the mathematical background computations used in the applica- tion, the actual implementation, testing meth- ods used and the results found, validating its educational efficacy in engaging students with complex physics concepts through interactive games.
%K Educational application; Physics; Unity; OpenCL
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p130-136.pdf

%T VRduino Learning: Introduction in Arduino with Virtual Reality
%A Georgiana-Alexandra Bodea
%A Victor Ioan Bâcu
%P 137-143
%X Traditional methods of learning Arduino often encounter challenges such as component damage, limited interactivity, and lack of real-world simulation. To address these issues, several 2D/3D applications have been developed, aiming to enhance traditional learning approaches by safeguarding components from damage, thereby reducing risks and costs associated with practical experimentation. However, despite these applications being cost-effective, they offer limited hands-on experience and do not mitigate potential disruptions. In this paper, we introduce an application aimed at enriching Arduino learning through immersive experiences. This application allows users to actively engage in learning and provides teachers with tools to create customized lessons and projects using the available components.
%K Virtual Reality; eLearning; Arduino
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p137-143.pdf

%T Virtual reality application for learning geography with gesture interaction
%A Maria Vasilache
%A Adrian Sabou
%A Teodor-Traian Ștefănuț
%P 144-150
%X Geography class lessons should adapt to new technology to improve teaching and increase student motivation. To enhance teaching experience, teachers can start with a map or globe presentation, followed by images and videos about the topic. However, organizing the lesson can be time- consuming, causing students to struggle when studying alone, thus the solution is to create a VR application such as GeoJoc. This application will provide information about rivers and mountains and include a quiz to test understanding. The immersive 3D world in GeoJoc will represent information through images and videos, enhancing interactivity, allowing users to interact with objects in the game using their hands and to perform actions with hand gestures. The key aspect of the game is to increase kids' motivation, attention, and understanding.
%K Virtual Reality; Meta Quest 2; Mountains and river lesson; Quiz; Educational Application; Hand Gestures Recognition; Object interaction.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p144-150.pdf

%T Tomorrow’s Classroom: Virtual Reality Chemistry Lab
%A Alina Duca
%A Adrian Iftene
%P 151-156
%X In the context of immersive technologies, the introduction of Virtual Reality (VR) in education opens up new perspectives for learning and teaching, allowing students to explore and interact with intuitively teaching materials and providing innovative ways to experience and understand complex concepts such as chemical reactions, the anatomy of the human body, the evolution of biological species, outer space, geological processes or historical events. This application shows how advanced technologies like VR can revolutionize the way students discover and understand information, allowing them to explore theoretical concepts visually and interactively. Through VR, theoretical notions become tangible and applicable, facilitating experiential learning that stimulates students’ curiosity and active engagement. By creating a virtual laboratory, participants can directly experience the effects and dynamics of various chemical reactions without the risks associated with handling real substances. This approach not only increases student engagement and curiosity but also improves information retention and a deep understanding of chemical processes. Also, our solution shows how, by incorporating Artificial Intelligence algorithms in virtual environments, we can take the field of eLearning to a higher level.
%K Virtual Reality; User experience; eLearning; Game Development; Natural Language Processing; Artificial Intelligence; guidance
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p151-156.pdf

%T Exploring the role of web accessibility in inclusive education: a case study
%A Marian Pădure
%A Costin Pribeanu
%P 157-162
%X Equal access to education is an important objective of the Digital Agenda for Europe. To ensure accessibility of the public web, systematic evaluation and monitoring measures are needed. This paper reports on a case study that targeted the accessibility evaluation of a Romanian university website. The approach to evaluation is user-centered. An evaluation instrument that consists of two sets of questions has been administered to students with various kinds of disabilities. The results show that web accessibility issues affect not only visually impaired students but also students with hearing and motor disabilities.
%K accessibility evaluation, university websites, WCAG2
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p157-162.pdf

%T The Development Process of eLearning Application for Configurations of the Operating Room
%A Daria-Elena Melinte
%A Adrian Iftene
%P 163-170
%X The application of 2D and 3D modeling has expanded across various domains, encompassing interior and exterior design, eLearning games and applications, animations, and futuristic advertisements. This work explores the development of an eLearning application designed to evaluate bioengineering students and assist them in preparing an operating room using 3D models. The findings of the exploration have led to the development of an application characterized by a multi-role framework. Within this framework, educators are afforded the capability to manage multiple groups of students, as well as to design and administer assessments that evaluate students' knowledge. Concurrently, students have the opportunity to enroll in multiple groups and complete the assessments provided by their instructors. Furthermore, both educators and students are granted access to a "playground" feature, which serves as a training or learning environment. This feature allows users to visualize the operational environment they are working within from various perspectives, thereby enhancing their understanding and interaction with the content.
%K Medical application; 3D Models; eLearning
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p163-170.pdf

%T Predictors of student satisfaction - The role of social capital and teacher support
%A Costin Pribeanu
%A Gabriel Gorghiu
%A Elena-Ancuța Santi
%P 171-177
%X The pandemic generated by the COVID-19 virus caused many changes in the education system. On the one hand, universities developed the infrastructure and provided more learning materials in digital format. On the other hand, the lockdown restrictions stimulated students to create new abilities and skills. Online communication with teachers improved as well as online communication with colleagues. Student satisfaction is a key variable in the evaluation of academic service quality and depends on many factors, such as motivation, aspirations, self-efficacy, and learning support. The purpose of this work is to analyze the interplay between three antecedents of student satisfaction: student self-efficacy, social capital, and teacher support. The results showed that while student self-efficacy was the main predictor, the other two factors had both a direct and indirect effect on student satisfaction.
%K Student satisfaction, academic self-efficacy, social capital, teacher support, online learning.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p171-177.pdf

%T INTELLIT Version 2.0 – Advancements and Evaluation of an Enhanced Platform for the Modern Romanian Literature
%A Alin-Gabriel Dita
%A Mihaita Stroe
%A Laurentiu-Marian Neagu
%A Mihai Dascalu
%A Lucian Chisu
%P 178-183
%X Creating digital solutions for Romanian literature is essential to preserve Romanian cultural identity. The development of the INTELLIT1 platform (version 1.0) represented a significant first step in the digitalization process of Romanian literature while also providing a high-quality educational resource. Overall, the platform aims to present information in a clear and user-friendly manner for those interested in enriching their knowledge of Romanian literature. INTELLIT is based on two essential corpora published by the “G. Călinescu” Institute of Literary History of the Romanian Academy: the General Dictionary of Romanian Literature (DGLR) and the Chronology of Romanian Literary Life (CVLR). Additionally, the platform features work by 32 Romanian writers from the “Fundamental Works" Collection. This study introduces the updated version (2.0) of INTELLIT that considers an improved architecture with a rewritten backend that indexes information in an Elasticsearch server, enhanced workflows including semantic search based on sentence embeddings, and an optimized UI/UX in the form of a responsive web platform. A study on this version showed that users had a pleasant experience, selected the most attractive features of the INTELLIT platform, and provided suggestions for future developments.
%K Web application; User Experience; Romanian literature; INTELLIT; DGLR; CVLR, semantic search.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p178-183.pdf

%T Improvement of Software Diagnostics with Advanced Execution Logging
%A Andrei-Ionuţ Nicolaev
%A Crenguţa-Mădălina Puchianu
%P 184-190
%X In this paper we present the development process of the Execution Lens, a software system that solves the problems of monitoring and re-running the processes of another application, with the aim of improving its quality and enhancing bug solving efficiency. The system consists of two packages intended for applications developed using C#/.NET and a web application for viewing and managing data saved for monitoring. Packages are in use to save the data needed to be monitored and respectively to be able to re-execute flows of the host application. The system was tested from the point of view of usability by a group of developers with at least one year of experience in the field. The result of the evaluation was a good one, the stakeholders being satisfied with the Execution Lens, finding it useful in the development of their applications.
%K software engineering, software logging
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p184-190.pdf

%T Improved Maths Solving Abilities for Transformers Using Flow Engineering
%A Nicolae-Robert Lincă
%A Traian Rebedea
%P 192-197
%X In this paper, we propose a flow engineering-based approach for improving the generation of solutions for math problems in the case of general content generative Natural Language Processing (NLP) models such as OpenAI GPT-4 and Google Gemini. Flow engineering makes use of a mix of prompt engineering techniques and has been recently proposed as a solution to improve code generation for transformer models. To test the performance of our proposed approach we use a subset of one of the most challenging datasets for mathematical solving, MATH, for which top transformer models exhibit accuracy up to 70%. Additionally, we aim to create a framework for generating highly accurate solutions and automating the correctness testing process of solutions. Advancing AI research by improving GPT models in mathematical problem-solving can increase the robustness and versatility of AI. This advancement facilitates interdisciplinary research and promotes collaboration across different fields including the design of efficient chat interfaces, unlocking limitless potential applications.
%K Natural Language Processing; Chat Interfaces; Mathematical Reasoning; Transformers; Flow Engineering.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p192-197.pdf

%T Smart Eco Driving Assistant
%A Ion Cojocaru
%A Paul-Stefan Popescu
%A Marian Cristian Mihaescu
%P 198-205
%X This paper presents a system designed and implemented for eco-driving behaviour estimation using machine learning methods. The system is described from a high-level perspective, detailing the components and workflow, including data collection from an OBD-II adapter and smartphone sensors. The dataset used in this study was meticulously labelled and processed to train the machine learning model. We employed various clustering algorithms to automate the labelling process. The eco-driving estimation model operates in real-time, providing continuous feedback to the driver through notifications. Experimental results demonstrate the model's effectiveness in distinguishing eco-driving styles, with data metrics analysis confirming the robustness of the approach. This system aims to encourage more fuel-efficient driving habits by providing actionable insights to drivers.
%K Machine learning, eco-driving, dataset.
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p198-205.pdf

%T Snoring Analysis System for Automatic Diagnosis of Obstructive Sleep Apnea Syndrome during VR Sleep Therapy
%A Dragoș Datcu
%A Florin Stoicescu
%A Cătălin Ciobanu
%A Dorin-Mircea Popovici
%P 206-213
%X While Virtual Reality (VR) sleep applications are not directly related to sleep apnea detection, they demonstrate the potential for technology to influence sleep behaviors and environments. As research in this field progresses, it may be valuable to consider how immersive technologies could be integrated with sleep monitoring and diagnostic tools, potentially enhancing patient comfort and compliance in sleep studies. Obstructive Sleep Apnea (OSA) is a prevalent and serious sleep disorder affecting an estimated 936 million adults worldwide. By leveraging the high prevalence of snoring in OSA patients, this approach has the potential to significantly improve early diagnosis rates and, consequently, patient outcomes. This study proposes an innovative method for OSA detection through automatic snoring analysis by fine-tuning Wav2vec 2.0 speech model to support VR sleeping therapy, aiming to provide a more accessible and cost-effective alternative to traditional polysomnography. Additionally, three non-deep learning techniques are presented together with an ESP32S-based edge system prototype as support for VR sleeping therapy.
%K VR, Sleep Apnea; Audio processing
%W https://rochi.utcluj.ro/icusi/proceedings/articles/1/ICUSI_2024_p206-213.pdf