10th International Congress on Information and Communication Technology in concurrent with ICT Excellence Awards (ICICT 2025) will be held at London, United Kingdom | February 18 - 21 2025.
Sign up or log in to bookmark your favorites and sync them to your phone or calendar.
Authors - Debjani Mazumder, Jiaul H. Paik, Anupam Basu Abstract - The large volume of online educational materials makes it difficult for learners to find adequate resources for better learning. Understanding these materials relies on identifying key concepts essential for comprehension. Automatic concept extraction is an important task in educational data mining and is similar to keyphrase extraction in Natural Language Processing (NLP). This process helps identify key ideas, organize documents, and build an insightful learning path. We present a probabilistic approach for concept extraction. Candidate concepts are generated using Wikipedia anchor texts. We identify the necessary concepts based solely on the educational context of a particular document using a graph-based probabilistic model. Evaluation of our method on two datasets (namely, a Physics school textbook and Physics articles 3) outperforms existing unsupervised and supervised methods.
Authors - Erich Giusseppe Soto Parada, German A. Montoya, Carlos Lozano-Garzon Abstract - The Internet of Things (IoT) is a fast-developing technological domain that has seen remarkable expansion in recent years; however, the security of these devices is critical, particularly with Denial of Service (DoS) attacks, which seek to overload a service or device by using a machine to render the device inactive. In this sense, we propose two machine learning approaches: a Random Forest approach, which has an F1 score of 0.99985 and an inference time of 0.457026 seconds for almost 500,000 records, and another from XGBoost, with an F1 score of 0.998989 and an inference time of 0.325767 seconds for the same 500,000 records. According to the data set, the methodologies used, and their results, these models were the most suitable for addressing the security issues imposed by DoS attacks.
Authors - Wirote Jongchanachavawat, Nirumol Hirunwijitporn, Noppon Mingmuang, Pisit Plaikaew, Supachai Poumpong, Narongsak Wornplop, Pannawat Koonmee Abstract - The increasing prevalence of firearms poses significant challenges to public safety, particularly in high-risk environments such as schools, airports, and transportation hubs. This study explores the implementation and performance of YOLOv11, the latest advancement in the YOLO series, for handgun detection in static images, video streams, and real-time CCTV monitoring. By leveraging its transformer-based architecture and adaptive scene understanding, YOLOv11 achieves exceptional accuracy, low latency, and minimal false positives across diverse scenarios. The results demonstrate YOLOv11's superiority over previous iterations in precision, speed, and robustness, making it a reliable solution for real-time threat detection. This research underscores the potential of integrating YOLOv11 into modern surveillance systems to enhance public safety and crime prevention efforts.
Authors - Pei-Yi Hao Abstract - Most stock prediction models rely on classification or regression methods to forecast stock price trends or prices, with their primary goal being to enhance the fit between predicted results and actual values rather than directly identifying the best investment targets. Consequently, the stocks recommended by these models may not necessarily yield the optimal returns. In contrast, stock ranking prediction provides a more direct and effective approach to portfolio construction by forecasting the ranking sequence of stock returns (with higher-return stocks ranked higher). This process is referred to as stock selection. The key to stock selection lies in identifying stocks that are most likely to help investors generate profits. Since stock prediction involves different tasks such as classification, regression, and ranking, which exhibit significant interrelations, most deep learning algorithms tend to train these tasks independently, overlooking their correlations. However, these related tasks may share underlying knowledge, which should be jointly learned to maximize the utilization of the potential information behind each task. Support vector machines (SVMs) have demonstrated exceptional performance in multi-task learning and have achieved success in numerous practical applications. This paper proposes a novel multitask support vector machine capable of simultaneously learning classification, regression, and ranking models. By leveraging the correlations among these tasks, the proposed framework aims to improve the predictive performance of each individual task.
Authors - Paula Escudeiro, Marcia Campos Gouveia, Nuno Escudeiro Abstract - The modern era is characterized by technological advancements, societal changes, and a reassessment of long-held paradigms. Within this shifting landscape, approaches to teaching and learning assessment have undergone substantial transformation. Modern pedagogical practices focus on understanding how students learn rather than merely assessing what they learn. Evaluating progress in online courses requires continuous assessment strategies that uphold the same level of credibility as traditional, face-to-face evaluations. The integration of quantitative and qualitative models, along with self-assessment and peer assessment, is vital for ensuring robust and effective evaluation in online learning environments.
Authors - Ainhoa Osa-Sanchez, Paulina Carcamo Ibarra, Begonya Garcia Zapirain Abstract - Breast cancer remains one of the most diagnosed cancers and a leading cause of cancer-related mortality worldwide. Advances in predictive modeling have introduced innovative methods to improve breast cancer prognosis and recurrence prediction, particularly through the integration of clinical, radiomic, and temporal data. This study focuses on the application of advanced feature selection techniques and machine learning algorithms, including Random Forest, XGBoost, and Lasso Regression, to optimize the performance and interpretability of predictive models. Radiomic features, such as the median of intensity histogram and the difference entropy of the grey level co-occurrence matrix (GLCM), alongside clinical and temporal variables, were identified as key predictors of recurrence. Our findings underscore the potential of combining multimodal data with robust feature selection techniques to enhance personalized treatment strategies for HER2-positive breast cancer patients. Future research should address dataset generalizability and incorporate multi-omics data to further refine these predictive approaches.
Authors - Bryan S. Guevara, Jose Varela-Aldas, Viviana Moya, Daniel C. Gandolfo, Juan M. Toibero Abstract - This study presents an innovative methodology for the development and testing of a digital twin for an Unmanned Aerial Vehicle (UAV), effectively bridging the simulation-reality gap. The proposed approach integrates Model-in-the-Loop (MiL) and Hardware-in-the-Loop (HiL) testing, enabling a comprehensive evaluation of the UAV’s digital twin in simulated environments. Behavioral testing includes open-loop scenarios, baseline feedback controllers, and Model Predictive Control (MPC). The UAV’s dynamic model is simplified and rigorously validated through experimental verification, ensuring high fidelity and reliability. Furthermore, this approach facilitates the critical transition from simulation to real-world experimentation by providing a robust framework for evaluating UAV performance under realistic conditions. This methodology highlights the importance of experimental validation in replicating real-world scenarios, ensuring the robustness and accuracy of the digital twin.
Authors - Restyandito, Danny Sebastian, Audrianne Gunawan Abstract - Borobudur Temple, a 9th-century Mahayana Buddhist monument in Central Java, Indonesia, is renowned for its intricate design and historical significance, attracting millions of visitors annually. Generation Z, being digital natives, prefers interactive and personalized experiences, making the AR Budur application an ideal solution for navigating the vast and intricate Borobudur Temple complex. By leveraging Augmented Reality, AR Budur aims to provide an engaging and intuitive experience that aligns with the tech-savvy nature of Generation Z, enhancing their exploration of this culturally significant site. Based on the research results Generation Z, especially female high school and undergraduate students, are ideal candidates for the AR Budur application due to their tech-savviness and high smartphone usage. Their navigation preferences emphasize the need for combining digital tools with human assistance and features like street view and real-time updates. Usability testing of AR Budur showed an average SUS score of 70.357, indicating above-average usability but with room for improvement. Feedback highlighted issues like elevation discrepancies in 3D mapping, which have been addressed, and suggestions for adding information about Borobudur Temple's reliefs for future updates.
Authors - Kunihiko Takamatsu, Tetsuya Oishi, Takahiro Seki, Kahori Ogashiwa Abstract - This study investigates the transformation of higher education in the post-COVID-19 era, focusing on the integration of Information and Communication Technology (ICT) and data-driven education approaches in strategic planning processes. Through a comprehensive survey of Japanese universities (N=816), including national (52.3% response rate), public (42.2%), and private institutions (15.8%), we examined how universities are adapting their medium-term planning frameworks to address contemporary challenges. The research reveals that 98.4% of respondents recognize the importance of data utilization in strategic planning, while highlighting the need for a more sophisticated approach that combines traditional methods with Eduinformatics frameworks. Our findings demonstrate that the post-COVID-19 landscape demands strategic plans that effectively leverage both quantitative metrics and qualitative assessments, particularly in evaluating educational outcomes. The study identifies key challenges in implementing data-driven approaches and proposes a comprehensive model for strategic planning that integrates ICT capabilities with institutional research (IR) methodologies. This research contributes to the emerging field of Eduinformatics by providing empirical evidence for the development of adaptive, technology-enhanced planning frameworks in higher education, while acknowledging the need for flexible, institution-specific approaches to strategic management in the post-COVID-19 era.
Authors - Ashok Kumar, Sanjeev Patwa, Sunil Kumar Jangir Abstract - Insider threats are one of the toughest challenges in cybersecurity. Insider attacks can be particularly dangerous because they often go unnoticed and can lead to serious problems like data breaches, financial losses, and damage to a company’s reputation. This issue has become even more pressing in recent years with the rise of digital operations and remote work. Research-ers have shown how machine learning can help predict these insider threats. While supervised learning models have shown great accuracy in identifying threats in certain datasets, they face a major hurdle: there simply isn’t enough labeled data on insider threats. On the other hand, unsu-pervised learning methods can spot unusual behavior and reveal hidden threats, but they often produce false alarms. Deep learning techniques could potentially offer better accuracy, but they require a lot of computing power and large amounts of training data. There are also exciting new trends in the field, such as behavioral biometrics, hybrid models, and explainable AI. However, challenges like inconsistent evaluation metrics and the difficulty of applying these models across different organizations still exist. This review aims to bring together existing research and pin-point key areas that need more attention, providing a roadmap for future studies. By addressing issues like the need for standardized datasets, encouraging collaboration across different fields, and incorporating contextual data from organizations, this paper seeks to help future researchers create more effective and adaptable models for predicting insider threats.
Authors - Key Sabbathany Togiana Tambunan, Nurul Sukma Lestari, Tri Wiyana Abstract - This study aims to analyze the relationship between Perceived Usability, Perceived Information, Perceived Service interaction, Satisfaction, and Engagement among the associates and the organization to which they apply. The methodology in this research paper involved quantitative data collected from a sample of 200 associates working within luxurious five-star hotels in Jakarta that implement online application platforms in the job application process. The data processed utilized the SEMPLS. The findings have revealed that a higher level of Perceived Usability, clear Perceived Information, and Perceived Service Interaction unquestionably contributes to an increase in satisfaction and engagement among candidates. The novel perspective of this study contributes by associating Perceived Information and Perceived Service Interaction directly to user engagement in Jakarta’s luxury hotels, by offering an insight that management is encouraged to prosecute during the job application process. This study discovers the new finding of the need for change to a wider range of comprehension of how clear communication and relevant information mold the user experience. Focusing on the experience of users in the application platforms, this research provides a unique understanding for management to increase the success of recruitment and nurture an engaging work environment for potential associates. Future studies are encouraged to explore these dynamics through extensive research to achieve a better understanding of their long-term impact on employee performance and retention in the fast-changing hospitality industry.
Authors - Xuening Tang Abstract - Human-Machine Communication (HMC) explores the joint meaning-making processes between humans and communicative intelligent machines, as well as its broader societal implications. The rapid advancements in Natural Language Processing (NLP) and Natural Language Generation (NLG) have transformed intelligent machines from passive tools to active social actors, capable of understanding and generating human-like messages. This paper evaluates the evolving research agenda of HMC, focusing on popular communication frameworks such as the CASA paradigm, the dual-process model, and social presence. It examines their theoretical foundations, transformations, limitations, and potential future extensions. Furthermore, this paper discusses the societal and ethical dimensions of HMC and provides recommendations for advancing research, including interdisciplinary approaches and the refinement of methodologies.
Authors - Dominic Gibietz, Daniel Helmer, Eicke Godehardt, Heiko Hinkelmann, Thomas Hollstein Abstract - The ongoing demographic change towards an aging population increases the need for effective solutions to support independent living and ensure the safety of elderly people living alone. Detecting anomalies in the daily routines of these people is a critical task in order to address these challenges and maintain their well-being. This paper proposes an unobtrusive method for anomaly detection using binary sensor data and machine learning. The approach involves a neural network in form of an autoencoder, which evaluates hourly data of each room, including the accumulated residence time, the activity time, and the number of room entries. The system learns individual normal behaviour through online learning and detects deviations from it. Testing and evaluation of the system was carried out using a publicly available dataset and comparing different configurations for the model. A comparison was also made between the use of individual maximum values for each room to normalize the data and uniform values for all rooms, with the former performing significantly better. The results demonstrate that the system can effectively identify the majority of unusual daily routines with a high accuracy, offering potential for improving safety measures for people living alone.
Authors - Chi-Hung Wang, Xiang-Shun Yang, Jun-Yi Liu, Yao-Jun Liu Abstract - Contract review is a common challenge for governments, businesses, and individuals. It becomes challenging when manual reviews are slow, legal expertise is lacking, and clauses are complex. These issues often lead to legal disputes and business conflicts. Traditional rule-based contract review tools often struggle with ambiguous language and unstructured content. Large language models (LLM) can quickly analyze contracts and find risks. But, they are unreliable due to "hallucinations" and a lack of knowledge of rare clauses. This study used retrieval-augmented generation (RAG) technology to overcome these challenges. It integrated verified legal data with large language models. This improved review accuracy to 93.67%. The F1-scores reached 91.95% for compliant clauses and 94.79% for non-compliant ones. The ROC-AUC metric improved to 0.93. The results show that this approach works. It improves the classification and risk identification of contract clauses. It also helps in contract review in the legal and business sectors, promoting the use of legal tech.
Authors - Samrat Ray, Souvik Datta, Smita Mehendale, Mita Mehta Abstract - The use of big data in social justice has become a phenomenon that is transforming the entire society, given that it provides solutions to challenges facing the world through the betterment of the lives of the affected groups of people. This paper focuses on the role of positive change by means of data science with a special emphasis on real-time data analysis in supporting power to the people efforts. It starts with the introduction of Data Science approaches and their connection with social transformation focusing on how it has made it possible for organizations to make sound decisions followed by the practical use of real-time big data to support research claims through the use of real-life case scenarios including poverty alleviation, city planning and development among others. Insights from these shed the light on ethical issues and need to make a conscious effort towards making data science solutions available for every segment of society. Finally, the paper analyses the trends and the future possibilities of data science for social enablement. It highlights the possibility of achieving even greater improvement of social programs through advanced research and development. In conclusion, this paper is a summary of how data science can be used to make society better which should prove useful as a reference for policymakers, researchers, and practitioners who are using data to initiate social change.
Authors - Chi-Hung Wang, Yu-Siang Siang, Yu-Hsuan Lin, Cheng-Hsien Lin Abstract - Aerial imagery is widely employed in intelligent transportation management and urban planning. However, dynamic objects often occlude critical information such as road signs and traffic markings, reducing the accuracy of image analysis and thereby affecting application reliability. Although traditional methods can partially address this issue, their high cost and low efficiency pose challenges in large-scale data processing. To overcome these limitations, this study proposes a background averaging technique based on real-time open-vocabulary object detection integrated with difference-based object detection using depth estimation. This approach enables zero-shot dynamic object removal, enhancing both processing efficiency and scalability. Experimental results demonstrate that our technique outperforms conventional methods across multiple performance metrics. Specifically, the multimodal framework combining depth-based differencing with the YOLO-world model achieves Precision, Recall, and F1-Score of 0.9062, 1.0000, and 0.9508, respectively. Furthermore, the Structural Similarity Index (SSIM) for background reconstruction reaches 0.9603, exceeding that of traditional YOLO models (SSIM = 0.9375). These findings indicate that our method not only effectively removes dynamic objects but also accurately restores background information, providing robust support for applications in intelligent transportation management and urban planning.
Authors - Wa Ode Siti Nur Alam, Riri Fitri Sari Abstract - The rapid development of artificial intelligence has facilitated the creation of Chatbot AI systems capable of addressing diverse healthcare challenges, including public education on critical issues like stunting. Leveraging Generative Pre-Trained Transformer (GPT) models and ensemble learning methods, such systems provide accurate, bilingual responses while ensuring scalability. A key implementation, deploying a Bilingual Chatbot AI through the Telegram application, demonstrates the feasibility of using accessible platforms to disseminate vital healthcare information. However, AI chatbots often face limitations, such as inaccurate or delayed responses, hindering user satisfaction and trust. Challenges in stunting education and nutritional status classification include adapting to linguistic nuances and ensuring real-time interaction. Addressing these gaps, we developed a GPT-Ensemble Learning-based chatbot to deliver information about stunting, including its definition, symptoms, impacts, prevention measures, and classification of toddlers' nutritional status based on gender, age, and height. The chatbots provide relevant responses for stunting education and nutritional status classification in Indonesian and English contexts. Our experiments also highlight Random Forest as the optimal ensemble model, achieving exceptional performance metrics: accuracy (0.99), precision (0.99), recall 0.96, F1-score (0.99), and ROC-AUC (0.99). This high performance ensures reliable nutritional status classification while improving accuracy and speed in bilingual interactions. The results underscore the potential of integrating AI-driven solutions into accessible applications like Telegram, which has significant implications for improving public health awareness and decision-making.
Authors - Andrea Augello, Alessandra De Paola, Domenico Giosue, Giuseppe Lo Re Abstract - Smart systems have become increasingly popular in recent years, widening the attack surface of cyber threats. Machine learning algorithms have been successfully integrated into modern security mechanisms to detect such attacks. Internet of Things (IoT) systems often have limited computational resources and are unable to execute entire machine learning pipelines. However, these systems often produce and manage sensitive data. Thus, it is preferable to avoid exposing their data to external analysis, e.g., on cloud systems. This work introduces SF-AE: a novel architecture that enables the distributed training of an anomaly-based intrusion detection system on devices with limited computational resources without exposing sensitive data. Experimental results on multiple datasets show that SF-AE outperforms state-of-the-art methods in terms of attack detection performance, at lower computation and communication costs for the participating devices.
Authors - Juanjo Mena, Juan Miguel Lorite, Antonio Patrocinio-Braz, Adrian Fernandez Abstract - In recent times, society has been influenced by technological advancements that have facilitated progress and brought corresponding modifications across various fields and environments, both academic and professional. Within this context, emerging technologies such as augmented reality (AR), virtual reality (VR), extended reality (XR), and the Internet of Things (IoT) have gained prominence. These technologies have significantly contributed to the improvement of diverse areas, including education, industry, and medicine, among others. In this regard, Active Triangle Kids was developed as a project based on augmented reality, specifically designed for children aged 3 to 6 years. The project encompassed the design, planning, and training of an optical recognition application, along with the creation of a demo for a video game utilizing augmented reality. As a result, each component of the project was successfully developed independently, ensuring effective training and programming. The project concludes by highlighting its unique aspects, identifying the current limitations of its components, and outlining potential future directions for further development and improvement.
Authors - Sinan Bicer, Abdulrahman Nasser Abass Abdo, Habib Dogan, Abdullah Genc Abstract - In this work, using SRR meta-resonators, a band-stopping waveguide filter (WGF) in C band (4-7.5 GHz) is designed and fabricated by using both CNC milling and SLA methods, and the effect of the fabrication methods on the filter performance is experimentally evaluated. The filter order for each case is increased from 1 to 7 and meta-resonators are used as many as the number of filter degrees. To determine the performance of the WGFs, some results such as frequency response, center frequency, fractional bandwidth (FBW), and quality factor (Q) values are given comparatively for each filter order. Also, the simulated and measurement results are in good agreement with each other. The measured results show that the performance of the WGFs fabricated by the CNC milling method is partially better than the filter fabricated by the SLA method. This decrease in SLA performance is thought to be due to the production methods. However, The WGF with the SLA method is nearly 50% lighter in weight than that produced with the CNC method. As a result, the SLA fabrication method is experimentally demonstrated to be a good alternative to conventional fabrication methods such as CNC milling.
Authors - Biswadeep Sarkar, Abdul Shahid Abstract - Stock market prediction remains a critical area of research due to its significant economic implications and inherent complexity. With advancements in machine learning, research interest has grown substantially in understanding the impact of textual data on financial forecasting. This study presents a hybrid FinBERT-LSTM model that combines sentiment analysis of quarterly earnings conference calls with traditional price prediction methods. We evaluate our model’s effectiveness against standalone LSTM approaches across six major US stocks from the financial and technology sectors. Experimental results demonstrate that the sentiment-enhanced hybrid model achieves superior predictive accuracy for four of the six studied stocks, as measured by Mean Squared Error (MSE), Mean Absolute Percentage Error (MAPE), and Accuracy metrics. Most notably, Citibank and Meta demonstrated substantial improvements when incorporating sentiment analysis, with MSE scores approximately 38 percent lower compared to predictions without sentiment data. Our findings contribute to the growing body of research on textual analysis in financial forecasting, offering practical implications for investment decision-making and aligning with the United Nations Sustainable Development Goal (SDG) 9 – Industry, Innovation, and Infrastructure.
Authors - Italo Santos, Jugurta Montalvao, Luiz Miranda Abstract - The class representation capacity in signal spaces spanned by arrays of metal oxide sensors (e-noses) is studied in this work. It is addressed in one of its simplest configurations, with a commercial MOX sensor running in two different temperature modulations, working as two different sensors. The class representation capacity of such an array is studied in the information theory framework. It is shown that, for steady-state measurements without drift, only a few tens of classes can be properly accommodated in the corresponding signal space, under moderate levels of noise.
Authors - Elvin Eziama, Remigius Chidiebere Diovu, Gerald Onwujekwe, Jacob Kapita, Victor L.Y. Jegede, Jegede T.T. Jegede, Solomon G. Olumba, Harrison Edokpolor, Adeleye Olaniyan, Paul A. Orenuga, Anthony C. Ikekwere, Emmanuel A. Ikekwere, Uchechukwu Okonkwo, Egwuatu C.A. Egwuatu, Charles Anyim, Jacob A. Alebiosu, Victor N. Mbogu, Benjamin O. Enobakhare, Toheeb A. Oladimeji, Anthony Junior Odigie, Adeleye Olufemi Abstract - By improving reliable communication between cellular vehicle-to-everything (CV2X), intelligent transportation systems (ITS) have the potential to revolutionize the real-time transportation sector. However, one element that hinders the seamless deployment of ITS is security issues. Resource limitations, anomaly types, false positives, and sensor interference are among the difficulties. Discrete Wavelet-Based Deep Reinforcement Learning with Double Q Learning (DWT-DDQN), a robust hybrid approach that combines the strengths of both discrete wavelet transform (DWT) and Double Deep Q Network (DDQN), is presented in the paper as an integrated mechanism that addresses the majority of these issues. It can dynamically adapt to the network, enhancing the Connected and Automated Vehicles (CAV) system’s safety and dependability. The dynamic approach is achieved by incorporating both the filtering and detection processes, which give a more robust and reliable performance output. Our numerical results clearly demonstrate the superior performance of DWT-DDQN over the existing conventional method at low and high levels of attack rates of α levels of 1% and 3%, and 5% and 7%.
