Food spoilage in vending carts leads to economic losses and health risks, particularly in resource-constrained environments. This study proposes a Federated Learning-driven IoT System to enable automated freshness monitoring in vending carts while preserving data privacy. The research investigates five key aspects: (1) the role of IoT sensors in environmental monitoring, (2) the effectiveness of machine learning models in freshness classification, (3) the capability of federated learning in ensuring data privacy, (4) comparative performance of federated learning techniques, and (5) the impact of ensemble methods on system robustness. Data from 2020 to 2023, collected from vending carts equipped with IoT sensors, were analyzed using machine learning and federated learning frameworks. The findings demonstrate that advanced IoT sensor integration improves environmental monitoring accuracy, ensemble-based machine learning models enhance freshness classification, and federated learning effectively maintains data privacy. Furthermore, stacking ensemble methods outperform other federated learning techniques, providing higher accuracy and robustness. The results emphasize the potential of federated learning in optimizing freshness monitoring while ensuring privacy and reliability, addressing critical gaps in food safety and IoT-based monitoring systems.
The increasing use of AI-generated text has introduced new challenges in detecting paraphrased content, particularly in lesser-resourced languages. This study investigates the effectiveness of different algorithms in identifying ChatGPT-paraphrased texts, focusing on the impact of word unigram and character multigram features, classification algorithm performance across English and Serbian corpora, and the comparative efficiency of commercial detectors like ZeroGPT against custom models. Additionally, it examines the role of syntax analysis and model temperature in influencing AI-generated text structures. A quantitative methodology involving classification algorithms, feature set evaluations, and cross-linguistic comparisons is employed. Results indicate that tailored algorithms outperform commercial detectors, especially when incorporating syntactic features. The study underscores the necessity for language-specific approaches to enhance detection accuracy and proposes directions for future research in AI text detection.
This paper explores the integration of emoji analysis into text-based emotion detection, emphasizing the significance of multimodal fusion in online communication. With the increasing use of emojis as emotional cues, understanding their impact on sentiment classification is crucial. The study investigates five key areas: the effect of emoji usage on emotion detection accuracy, the role of emojis in differentiating supportive and contrastive sentiments, the impact of emoji context on sarcasm interpretation, the integration of emojis in hybrid deep learning frameworks, and the effectiveness of multimodal fusion techniques in enhancing emotion classification. Using a quantitative research approach, this study leverages the GoEmotions dataset to analyze the relationship between emoji usage and emotion detection performance. Findings demonstrate that incorporating emojis significantly improves classification accuracy, sentiment differentiation, and sarcasm interpretation. Additionally, hybrid frameworks integrating emojis enhance emotion detection capabilities, and multimodal fusion techniques improve classification performance. The research contributes to the growing field of emotion detection by highlighting the essential role of emojis in enriching sentiment analysis models. Future work should address dataset diversity and cultural factors to refine emotion detection frameworks further.
The rapid advancement of genomics data analysis on High-Performance Computing (HPC) cloud infrastructures presents unique challenges in ensuring compliance with the General Data Protection Regulation (GDPR). This study examines GDPR-compliant solutions for large-scale genomics data analysis, focusing on the strategies employed by the CINECA supercomputing center. Key research questions explore data security, compliance measures, efficient data processing, integration of diverse diagnostic data, and scalability across institutions. A quantitative research approach is employed, analyzing relationships between data security measures and compliance rates. Findings indicate that dynamic GDPR-compliant protocols, advanced encryption, optimized data processing, standardized integration processes, and adaptable frameworks significantly enhance security, efficiency, and compliance. The study contributes to the field by bridging gaps in scalable and secure genomics data processing, offering insights into sustainable GDPR-compliant implementations.
Hate speech detection on social media has become a critical issue, particularly in bilingual settings where language mixing complicates identification. This study focuses on Amharic and Afaan Oromo, two widely spoken languages in Ethiopia, and investigates how deep learning techniques can enhance bilingual hate speech detection. The research examines five key aspects: the impact of language mixing on detection accuracy, the effectiveness of hybrid deep learning classifiers, the role of feature extraction techniques, the significance of linguistic features, and the influence of bilingual communication on hate speech propagation. Using classifiers such as CNN, BiLSTM, CNN-BiLSTM, and BiGRU, along with feature extraction methods like Keras word embedding, word2vec, and FastText, the study demonstrates that hybrid models outperform conventional approaches. The findings reveal that language mixing reduces detection accuracy, while advanced feature extraction techniques and linguistic feature integration significantly improve performance. The results contribute to addressing gaps in existing literature and provide insights into optimizing bilingual hate speech detection models. Future research should explore real-time detection methods and broader linguistic applications to enhance hate speech mitigation strategies on social media.