%T  International Conference on Human-Computer Interaction - RoCHI 2021
%C  Bucharest
%D  16-17 September 2021
%G  ISSN 2501-9422
%E  Mihai Dascalu
%E  Suzanne Kieffer
%P  190
%W http://rochi2021.utcluj.ro

%T Gamification, Persuasive Techniques, and Nudging: What is the Impact on the User Experience?
%A Olga De Troyer
%P 1-3
%X An important challenge faced by many software designers is how they can make users “hooked” to their software, i.e., what can they do to ensure that users continue to use their application over time, especially when there is no obligation to do so. Many apps suffer from the novelty effect and are abandoned quickly. Therefore, there is an increased interest in existing methods, techniques, and theories from different fields (such as psychology, social science, games) that can be used to keep users motivated to use the software. Nowadays, most popular are gamification, persuasive techniques, and nudging, all having their own characteristics and merits. We investigate these techniques from a user experience perspective. When can they be used successfully, when can they backfire? How can they be incorporated into software and are there ethical considerations? This paper is an extended abstract of an invited talk given at RoCHI 2021.
%K Gamification; Persuasion; Nudging; User Experience; Personalization; Personality types, Gamification User Types.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-DeTroyer.pdf

%T Challenges and Implications of Measuring User Experience for Wellbeing Research
%A Effie Lai-Chong Law
%P 4-6
%X The notion of User Experience (UX) is increasingly applied to wellbeing-driven AI-infused technology. Despite the trend, approaches to measuring UX still draw heavily on the conceptual and methodological frameworks of emotions; a variety of self-reported scales and psychophysiological tools exist. Chatbots are used as an example to illustrate the challenges pertaining to human-AI UX; implications for addressing them are inferred.
%K User Experience, Measurability, Emotions, Wellbeing, AI
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Law.pdf

%T Conversational Agent Embodying a Historical Figure using Transformers
%A Alexandru Dima
%A Traian Eugen Rebedea
%P 7-14
%X In this paper, we discuss how transformers, the current state of the art solution for Question Answering (QA) can be applied for implementing a conversational agent that can offer accurate and informative answers to questions about any user-chosen historical figure. We use Wikipedia to obtain data for every personality chosen and an information retrieval engine to store the data and perform full-text queries for obtaining the context dynamically. We tested the performance and accuracy for the different transformer models, data store, and retrieval methods. The best result was obtained using an ALBERT model with a heuristic retrieval method. This solution obtained an F1 score of 66.06 and an Exact Match (EM) score of 42.22.
%K Conversational agent; natural language processing; question answering; data retrieval; transformers.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Dima.pdf

%T Analysis and prediction of politeness in conversations
%A Cristiana-Paula Iordache
%A Ștefan Trăușan-Matu
%P 15-20
%X This paper describes an approach for the analysis of the politeness level in a conversation and a prediction model measuring the probability that a conversation will evolve into a conflict between its participants. A sliding window method is used for analyzing the most recent utterances in a conversation, in order to assign a politeness score for the conversation in every moment of its development. The approach presented in the paper enhances the analysis by using a punctuation tuning method for each utterance in the sliding window and by including a polyphonic analysis in the conversation score computation. A graphical interface model is introduced for empowering the user to conduct a detailed politeness analysis of the development and evolution of conversations. Potential uses of the analysis are described, with an emphasis on chat moderator bots capable of alleviating conflictual situations in a conversation, while also suggesting further improvements to the analysis model.
%K Conversation analysis; analysis of politeness; prediction of politeness; polyphony analysis;
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Iordache.pdf

%T Detecting deception in conversations
%A Ioana-Raluca Zaman
%A Ștefan Trăușan-Matu
%P 21-27
%X Interaction with other people is part of anyone’s life, and implicitly comes with the risk of deception. On average, a person is lied up to several dozen times a day, the main purpose being manipulation. Deception is defined as a psychological act of misleading or trying to persuade someone to believe certain facts. One manner to understand deception is via emotions and sentiments of the participants in conversation. This paper’s aim is to combine the information extracted from the text with the help of techniques specific to Natural Language Processing with notions of psychology. Even if lies are not a mandatory element, in most cases they play a crucial role in manipulation. Therefore, this project classifies lies based on the emotions extracted from a message, but also from the replies to it and using two alternatives of the Naïve Bayes algorithm. Moreover, the project will focus both on actual lies and suspected lies. The corpus utilized for analyze is Deception in Diplomacy Dataset from Cornell University ConvoKit.
%K natural language processing; deception; chat; sentiments; emotions
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Zaman.pdf

%T An Automatic Car License Plate Recognition System
%A Ionuț-Andrei Pătrăuceanu
%A Adrian Iftene
%A Daniela Gîfu
%P 28-35
%X Taking into account the increase in the number of cars and, implicitly, the traffic problems, an automatic license plate recognition (ALPR) system becomes a significant task in smart surveillance and transportation. The interest for the implementation and integration of ALPR technology into the daily security operations is visible, although there are still insufficiently resolved issues. They are based on either the help of image processing techniques or deep learning techniques. More precisely, they are using object detection algorithms such as You Only Look Once (YOLO). In order to obtain a system capable of detecting the license plates, we propose a method using YOLOv4. This system achieved impressive frames per second, the results show that we are talking about a method that is relatively insensitive to background variations.
%K Image classification, OCR, YOLOv4, Usability tests.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Patrauceanu.pdf

%T Live It - Recommendation System based on Emotion Detection
%A Diana Isabela Crainic
%A Adrian Iftene
%P 36-43
%X This paper presents the development of a web application that integrates a system of movie recommendations using the collaborative filtering algorithm with a component of real- time recording and detection of emotions. So far, there were no implementations that combine recommendations and emotions, so this application proposes these two to work together to make our lives better regarding the movie- watching experience. To recognize emotions, we created a component that examines facial expressions, which offers, as a result, one of the emotion types: happy, sad, neutral, anger, surprise. This component was later integrated into a movie recommendation application, which analyzes the user’s emotions in real-time while watching the presentation video. In the second part of the paper, we presented how we performed usability tests in order to improve the quality of the application. The results were promising, with a high degree of accuracy and usefulness coming from end-users, showing the future potential of this application, for instance adding new functionalities or recommendation algorithms.
%K Face recognition; emotions identification; recommendation system.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Crainic.pdf

%T Terrain Generation with Continuous Level of Detail
%A Lucian-Valentin Deaconu
%A Constantin Ioan Nandra
%A Dorian Gorgan
%P 44-51
%X This paper describes the implementation of a terrain generation technique, featuring a solution meant to address the problem of continuous level of detail. The implemented technique is employed as part of a custom-built rendering engine that can be extended by programming-proficient users to create desktop applications with 3D rendering capabilities. Throughout this paper, we focus on the challenges posed by generating and rendering terrain with variable level of detail, dependent upon the camera position. We analyze the roots of the problem and provide a tessellation-based solution to solve it, while maintainig the accuracy of the topology. The description of the various implementation aspects will also include some insight into the basic data structures employed.
%K Dynamic Terrain Generation; Continous Level of Detail Problem; OpenGL.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Deaconu.pdf

%T TreasAR Hunt – Location-Based Treasure Hunting Application in Augmented Reality for Mobile Devices
%A Adina-Maria Oros
%A Adrian Sabou
%P 52-59
%X TreasAR Hunt is a mobile application that blends together Augmented Reality and location tracking methods. It addresses to both technical and non-technical users. The application comes to enhance the learning process by creating an immersive environment that keeps the user interested and engaged. Thanks to the type of game it models, a treasure hunt, and to the fact that it involves interaction between the user and its surroundings, the application can also be used in tourism, as a city guide. In addition to that, it can be seen as a team-building tool, that dares players to collaborate. This paper describes the project’s objectives, its use cases, core functionalities and development phases. User testing results are provided and interpreted at the end, so to conclude how does the application fit in the stated scopes.
%K Augmented Reality; Marker-based detection; Location Based Services; Location tracking; Unity 3D; Educational application; Training application;
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Oros-Adina.pdf

%T PlayChemy: Educational Application in Augmented Reality for Equating Redox Equations
%A Maria Nistor-Brana
%A Adrian Sabou
%P 60-67
%X Augmented reality combines the virtual world with the real world by superimposing various digital objects on top of real objects. This type of technology brings many benefits to education, including making it easier to accumulate new information or boosting memory capacity. We propose to develop an augmented reality-based game with the aim of facilitating learning and understanding of methods for determining oxidation numbers and equating redox equations. The paper presents the development steps that have been followed in the realization of this project, from the definition of the application context to the actual implementation.
%K Augmented reality; Educational application; Chemistry; Unity
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Nistor-Brana.pdf

%T The Astronaut Game – A Vocabulary Game Employing Natural Language Processing Techniques
%A Irina Toma
%A Bianca-Eugenia Stefan
%A Mihai  Dascălu
%A Ștefan Trăușan-Matu
%P 68-73
%X Serious games are viable alternatives for learning and for acquiring new skills. Serious games introduce users with a learning environment that mimics real life situations, where they can practice and improve their skills. While considering education, we focus on games used for language learning, which cover multiple categories that range from vocabulary acquisition, comprehension, collaboration, to free text writing. The serious game described in this paper – the Astronaut Game – is design for English as Second Level users and consists of two parts. The first part consists of a Platform Jumper for vocabulary acquisition, in which an astronaut runs and jumps to collect rocket pieces that symbolize words; the aim of this sub-game is to teach words from different lexical fields to the players, depending on the level they are playing. The second part, the Rocket Hangman, is centered on comprehension and users must identify the central keywords from the reference text. Our game was tested by a group of 16 individuals and received positive feedback.
%K Serious Games; Language Learning; Vocabulary Acquisition; Comprehension; ReaderBench framework
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Toma.pdf

%T NSMuseum – A Case Study for Developing Cross-platform Mobile Applications in Augmented Reality
%A Mircea-Ioan Dragotă
%A Adrian Sabou
%P 74-81
%X This paper presents a method to streamline the process of developing a mobile application in augmented reality, in order to maximize exposure to a wider audience. In addition, a comparative analysis will be developed between the best- known ways to develop cross-platform applications in augmented reality, ARFoundation and Vuforia. The process described in this paper aims to add value to such an application, integrating Unity with another cross-platform technology, Flutter. As an example of this development process, a cross-platform application was created that offers an appearance as close as possible to the native one and an experience in augmented reality as pleasant as possible, being accompanied by a modern user interface. In addition to this, the example also focused on the integration of augmented reality in the museum visiting experience, enhancing the educational process. Our cross-platform approach has allowed the creation of a single application running on both iOS and Android operating systems, so both the cost, development time and resources have been reduced.
%K Augmented Reality; Cross-platform; Flutter; Unity; Mobile Application;
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Dragota.pdf

%T Usability evaluation methodology for scalable multiuser games
%A Bogdan-Ioan Oros
%A Victor Bâcu
%A Dorian Gorgan
%P 82-89
%X The aim of this paper is to propose means for objective evaluation metrics in case of multiuser online games and to offer a viable solution with a scalable architecture for the coordination services for this specific type of games. This paper presents a covering analysis on the high usability of such a solution with justification based on objective usability metrics, reusable for different game designs and scenarios. The article justifies the need of switching to a set of proposed metrics as means to usability evaluation instead of relying strictly on user feedback based on empirical analysis of the game.
%K Usability; HCI; human computer interaction; scalable; computer games; multiuser; multiclient;
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Oros-Bogdan.pdf

%T Intelligent Avatar Interaction in 3D Computer Games
%A Miklos Balazs
%A Dorian Gorgan
%P 90-97
%X The artificial intelligence elements make the computer game more unpredictable and attractive by increasing the creative and emotional involvement of the players. An avatar can be directly controlled by the user through interaction techniques or at certain times, the avatar can make decisions alone in a behavior not controlled by the user. This paper attempts to determine experimentally how the intelligent characters may be trained, what are the level of intelligent behavior an avatar can reach through learning, or how the avatar makes decisions on his own through an adaptive behavior. The paper highlights solutions for efficient and successful training of the avatar.
%K Intelligent agent, computer game, reinforcement learning, user interaction, avatar.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Balazs.pdf

%T Reinforcement Learning Agent for a Flight Simulation Video Game
%A Victor Nonea
%A Radu Iacob
%A Traian Eugen Rebedea
%P 98-102
%X This paper is a case study analysis of the viability of using machine learning methods for skilled NPC agents in production level video games and how they compare to their hand-coded counterparts. The implementation and experiments were made in a simple OpenGL game about flying an aircraft through a set of checkpoints without crashing. Our conclusion is that current machine learning methods are not feasible for building the NPC agent, because, while they simplify the agent’s design, they exponentially complicate the testing and debugging processes without offering an improved ability for the NPC.
%K Reinforcement learning; REINFORCE algorithm; video game; non-playable character; NPC; flight simulation.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Nonea.pdf

%T Early Usability Evaluation to Enhance User Interfaces - A Use Case on the Yggdrasil Cybersecurity Mockup
%A Denis Iorga
%A Octavian Grigorescu
%A Mihai Predoiu
%A Cristian Sandescu
%A Mihai  Dascălu
%A Razvan Rughinis
%P 103-110
%X Most usability evaluation methods are designed to be applied after the implementation of a software product. As such, there is limited research on how to conduct usability evaluation procedures before the implementation of a software product, especially on cybersecurity platforms user interfaces. For this purpose, we illustrate the early usability evaluation of a cybersecurity graphical interface. In this sense, we conduct a usability evaluation procedure on the mockup of Yggdrasill, an application centered on the early identification of vulnerabilities using open-source intelligence, more specifically Tweeter posts and news websites. To achieve this, we conducted a usability inquiry procedure using the adaptation of a questionnaire designed to evaluate the attractiveness of interactive products. In addition, we conducted a usability testing procedure based on the think-aloud protocol. The results of this study enabled the timely identification of improvement opportunities for the graphical interface of the Yggdrasil graphical interface before the actual implementation.
%K Early usability evaluation; Open-Source Intelligence; Graphical User Interface; Cyber-security.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Iorga.pdf

%T Design-Thinking and UCD Combination for Designing Effective Time Management Assistant Mobile App
%A Liyanage Ashana
%A Siriwardana Samanthi
%A Mithsara Malithi
%A Reyal Shyam
%P 111-118
%X  This project presents the high-fidelity prototypes of a smart digital personal assistant for university students in the form of a mobile app, using a combination of User-Centered Interaction Design and Design Thinking methodologies. The end goal is to have an effective and user-friendly set of interfaces that display the right amount of information to the user without missing key details while not showing too much information such that the user becomes overwhelmed. The target audience benefitting from this work are university students who are reactive in nature and do not possess the required perseverance to complete the tedious initial setting up process of a Time Management Tool (TMT). The study was conducted involving university students using surveys, interviews and focus groups. This sheds light on key time management issues faced by university students and paves the way for future research and development of TMTs.
%K User Centered Interaction Design (UCID), Design Thinking, Time Management, Human Computer Interaction, Time Management Tool (TMT)
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Ashana.pdf

%T LiftYourself: A health promoting mobile application
%A Andreea Cojocaru
%A Elena Bautu
%P 119-124
%X With the aim of promoting a healthy life with respect to physical activity and nutrition, this paper presents an attractive mobile application that engages the user on three directions. It proposes readymade fitness training sessions that target specific body areas, with the possibility to customize them as the user sees fit. The app allows the user to keep track of the nutrients consumed daily with respect to an ideal ratio that fits the individual profile of the user. Finally, menus are proposed to the user. Interactivity is supported by the incorporation of a speech based chatbot, developed using the IBM Watson technology. A questionnaire based usability study was conducted and measured the satisfaction level of the users that used the app, as well as gathered suggestions for improvement. This application has the potential to be extended for a wide range of uses, including clinical, preventive, public health or rehabilitation.
%K Android; Mobile app; Health.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Cojocaru.pdf

%T Spatial audio music player for web
%A Leischner Vojtěch
%A Míkovec Zdeněk
%P 125-129
%X Spatial audio offers new expressive means to music composers. Spatial audio technology is well supported in web browsers for desktop and mobiles platforms alike. However, the spatial audio is only marginally represented in web music industry. Most of the online streaming platforms such as Spotify, YouTube or Soundcloud use predominantly stereo format for playback. We propose a new easy to use web- based spatial audio player. We also propose spatial audio encoding using existing audio codecs and discuss the practical implementation. Evaluation of the player was conducted using qualitative user research and by logging measurable data during listening. We aim to democratize the use of spatial audio on the web for non-expert users.
%K Web audio; spatial audio; music; web design; streaming
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Leischner.pdf

%T Generating music from poems using neural networks
%A Horia-Ioan Iancu
%A Ștefan Trăușan-Matu
%P 130-133
%X Sonification is the process of using sound to convey or add a new dimension to information. In particular, textual human- computer interaction is enhanced if sound is added. In order to sonify a poem, a quantified representation of emotions in the form of sentiment dictionaries would be necessary, as well as knowledge of music theory. Diverging from previous research, a method that uses deep neural networks to combat both of these problems is presented. In order to establish an emotional connection between text and music, the general sentiment of a poem is first extracted. An analysis of the poem’s musicality, expressed in the form of rhythm, rhyme and punctuation, is then used as the basis for creating a simple rule-based melody. Using this melody as conditioning, a piano accompaniment is then produced using a neural network. Results show that the accompaniments generated adequately reflect the mood of a poem, while also being pleasant to listen to.
%K Sonification; poetry; Natural Language Processing; Neural Networks.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Iancu.pdf

%T An Augmented Reality Piano Learning Tool
%A Andra Simion
%A Adrian Iftene
%A Daniela Gîfu
%P 134-141
%X Music learning (here, thinking of playing piano) without professor can be a real challenge. This can be difficult to achieve, given some traditional approaches to manage this process are still insufficiently satisfactory. Beyond the well- known computer-assisted music learning tools (1) screen- based and (2) hardware-based, a promising solution for piano learning is based on augmented reality (AR). This technology provides an ideal environment for the student, but of course, it is not enough to become a career musician. In this paper, we propose an application that helps all those who want to learn to play the piano in a friendly way, starting with understanding sheet music. By projecting information onto the instrument, our AR system can create more direct interactions for the user.
%K ELearning, Augmented reality, Playing piano.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Simion.pdf

%T Exploring the factors making online learning attractive and enjoyable
%A Costin Pribeanu
%A Gabriel Gorghiu
%A Elena Ancuța Santi
%A Valentina Iuliana Manea
%A Tudor Macavei
%P 142-146
%X The pandemic forced the universities to go online and brought in front the online lectures as focal activities. A lot of effort has been put to move the teaching and learning resources on the online platform which became the main interaction space between students and between students and teachers. An exclusive online learning environment has both benefits and disadvantages, especially as regards the students’ engagement with the online lectures. Maintaining students’ interest is difficult and depends on several factors. The purpose of the paper is to explore the factors that make online learning attractive and enjoyable. A formatively measured model having as focal construct the intrinsic motivation has been tested on a sample of 354 Romanian university students. The results highlight the students’ need for self-presentation and show that social learning features are key factors that make online learning attractive and enjoyable.
%K Online learning, online lectures, online platform, learning motivation, intrinsic motivation.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Pribeanu.pdf

%T Extrinsic and intrinsic motivation in online education during the pandemic as perceived by Lithuanian and Romanian university students
%A Vincentas Lamanauskas
%A Rita Makarskaitė-Petkevičienė
%A Gabriel Gorghiu
%A Valentina Iuliana Manea
%A Costin Pribeanu
%P 147-154
%X The pandemic generated by the Coronavirus COVID-19 challenged universities to go exclusively online thus affecting both the education and personal life of teachers and students. This work analyzes the perceptions of Lithuanian and Romanian university students as regards the extrinsic and intrinsic motivation of online learning during the pandemic. Following the technology acceptance theory, extrinsic motivation has been operationalized as the perceived usefulness and intrinsic motivation as the perceived enjoyment. Although the results show a low to moderate motivation for online/distance learning, interesting differences between the two motivating factors have been found, as well as between the perceptions of students from the two countries.
%K Online education, online lectures, online educational platform, intrinsic motivation, extrinsic motivation, TAM, pandemic.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Lamanauskas.pdf

%T Development and Usability Evaluation of a Virtual Reality Application for Exploring University Labs
%A Radu Stamatoiu
%A Silviu Vert
%A Radu Vasiu
%P 155-162
%X Although virtual reality (VR) was initially created as an entertainment medium way back over 70 years ago, it has developed beyond its scope and we are now using it for many other purposes, like education, training and medicine. When it comes to an amazing VR experience, what glues everything together is the level of interactivity and the usability of the application itself. In this paper, we describe a mix of in-person and remote usability testing sessions, with analytics and questionnaires, that we performed to evaluate an educational VR application which was built for Politehnica University of Timisoara (Romania). We conclude with some general feedback and insights that might help other researchers and practitioners in improving similar applications.
%K Virtual reality; usability testing; educational application.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Stamatoiu.pdf

%T MoleCool: Augmented Reality in Chemistry Education, Possibilities and Solutions
%A Eszter Lokodi
%A Teodor Ștefănuț
%P 163-168
%X Chemistry is a very complex and difficult field to understand in the natural sciences. We all know the popular expression: everything around us is chemistry. Any object we see or touch is chemistry, the food we eat is chemistry, the air we breathe is chemistry, the emotions we experience are chemistry, our body is chemistry. But little do we know or understand how this actually manifests in reality. The current research presents an original solution for chemistry learning, based on Augmented Reality and Gamification principles. The main purpose of this approach is to help students visualize the microscopic molecule structures and describe them in an interactive manner. After a short overview of Augmented Reality initiatives in education the focus switches to the chemistry related initiatives. Finally, the details of the proposal are being discussed.
%K Augmented Reality; Chemistry education; Mobile applications.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Lokodi.pdf

%T Fake reviews detection techniques
%A Ioana-Ruxandra Stăncioi
%A Ștefan Trăușan-Matu
%P 169-175
%X In a world of a rapidly increasing e-commerce, fake reviews are a current problem for the customers and businesses. This thesis addresses the detection of fake reviews, solutions and improvements using machine learning techniques. Manual detection of fake reviews is a tedious process, which requires recognition of manipulative writing and examination of the questionable user’s profile.
%K Reviews; fake; ARI; sentiment; Wald-Wolfowitz.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Stancioi.pdf

%T Lib2Life – Domain Categorization of Books using BERT Language Models and Knowledge Graph Population
%A Melania Nițu
%A Mihai  Dascălu
%A Gabriel Guțu-Robu
%A Maria-Iuliana Dascălu
%A Silvia Tomescu
%P 176-181
%X Ontologies provide domain conceptualizations for storing knowledge in a unified and structured machine-readable format. Following the initial conceptualization of the domain centered on defining classes and corresponding properties, the process of properly instantiating and keeping up-to-date individuals can become tedious. Thus, the need for knowledge processing systems that automatically populate and update their knowledge representations increases. This paper introduces an approach to manage, organize, and automatically populate the Lib2Life ontology containing historical documents provided by the Central University Libraries in Romania. Our method integrated in the Lib2Life platform instantiates individuals in the RDF knowledge graph following book categories that are inferred using state- of-the-art language models. As such, new documents to be included in the Lib2Life platform are automatically categorized with an F1-score of .73 using a fine-tuned Romanian BERT model.
%K Digital library; Language models; Domain classification; Ontology population.
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Nitu.pdf

%T Topic-based Models with Fact Checking for Fake News Identification
%A Vlad Cristian Dumitru
%A Traian Eugen Rebedea
%P 182-190
%X The phenomenon of fake news has expanded more and more over the last period of time. This expansion is strongly related to the evolution of technology and social networks, which facilitates the distribution of information more than ever. In the context of this phenomenon growing up so fast and bringing unjust influences amongst people, studies have emerged on the automatic solutions proposed to combat fake news in the online environment. The study started with collecting a dataset that also contains more current information. As a first step the paper proposes a division of news according to topics using clustering algorithms. Subsequently, the system involves the training of a detection model within each cluster obtained, as well as the training of a global model. The study examines the possibility of using these models to form a mixed detection system. In the second part of the study, the focus is on conducting experiments based on the facts that appear in the news and the possibility of filtering the articles according to the relevant facts using an information news related database.
%K Fake news detection, Linguistic features, Natural language processing, Topic-based model
%W http://rochi.utcluj.ro/articole/9/RoCHI2021-Dumitru.pdf