Volume 13 | Issue 7 |

 Abstract

The increasing prevalence of cyberbullying on social media has necessitated the development of advanced detection mechanisms. Machine learning (ML) and natural language processing (NLP) techniques provide an effective means to analyze vast amounts of text data and identify cyberbullying patterns. This paper explores the application of ML and NLP techniques in detecting cyberbullying behavior. The methodology involves preprocessing social media comments, extracting relevant linguistic features, and training classification models to distinguish between bullying and non-bullying content. Various machine learning algorithms, such as logistic regression, decision trees, random forest, gradient boosting, and K-nearest neighbors, are employed. The experimental results indicate that the random forest classifier outperforms other models in accuracy, demonstrating the efficacy of the proposed system in detecting cyberbullying. Additionally, the paper discusses challenges such as detecting sarcasm, handling multilingual text, and mitigating bias in training datasets. Future work involves enhancing model adaptability using transformer-based architectures and integrating explainable AI techniques for improved interpretability. Moreover, considerations for real-time deployment, ethical concerns, and user privacy are addressed to ensure responsible AI-driven moderation. The results highlight the potential for real-time applications and automated moderation tools.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Machine learning (ML), natural language processing (NLP), sentiment analysis, classification models, explainable AI, transformer models, real-time monitoring, ethical AI, automated moderation

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

An aptitude test generator is a software application designed to create, customize, and administer aptitude tests for various purposes, such as recruitment, academic assessments, and skill evaluations. This method creates questions on the fly in a variety of areas, including as verbal ability, numeric aptitude, logical reasoning, and domain-specific knowledge. An aptitude test generator is an automated system made to effectively develop, administer, and assess aptitude tests. The platform offers a dual-access system that allows students to take tests, check results, and monitor their progress, while administrators may create tests, alter question banks, establish difficulty levels, and analyse student performance. Randomisation, adaptive testing, and real-time evaluation are all incorporated into the system to guarantee a uniform and equitable evaluation procedure. The system's features, which include automatic grading, question shuffling, and comprehensive performance analytics, improve accuracy, lessen administrative burden, and guarantee an impartial and enjoyable testing experience for teachers and students.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

APTITUDE TEST GENERATOR

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Spam calls have become a widespread nuisance, leading to wasted time, privacy concerns, and potential financial scams. To address this issue, we present Callnsight, an automated spam call detection system that leverages speech-to-text conversion and natural language processing. The system processes audio input from phone calls, converts it into text using AWS Transcribe, and analyzes the transcript using Google Gemini API to determine whether the call is spam. The API's output, structured in JSON format, enables easy extraction of relevant insights for classification. Callnsight provides a scalable and efficient approach to spam detection, offering real-time analysis and improving user security. This paper details the system architecture, implementation process, and potential improvements for enhancing spam detection accuracy.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Spam call detection, speech-to-text, AWS Transcribe, Google Gemini API, natural language processing (NLP), call classification, JSON, automated spam filtering, AI-driven spam detection, real-time call analysis

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

In order to facilitate intuitive and touchless control of brightness and volume, this article focuses on the creative application of hand gesture recognition and emotion detection through facial recognition. The technology uses machine learning algorithms and sophisticated computer vision techniques to identify particular hand motions and dynamically change the screen's brightness and audio levels, offering a practical and effective substitute for conventional physical controls. Furthermore, by analyzing facial expressions and adjusting environmental settings--such as turning down the lights or volume when melancholy is detected or turning up the brightness and volume for happy moods--the system's incorporation of emotion recognition enables it to customize the user experience. The hands-free interface provided by this initiative, which emphasizes inclusivity and accessibility, can help people with disabilities or those in sterile settings where touchless contact is crucial. Through adaptive brightness adjustments, the technology optimizes energy utilization and further advances sustainability. In addition to improving user comfort and interaction, this study shows the potential for human-centric smart automation by fusing gesture recognition and emotional intelligence. This could lead to applications in home automation, healthcare, education, and entertainment.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Brightness, Volume, Detection, Emotion, OpenCV, Python, Facial, TensorFlow, MediaPipe

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Although modern technology has made significant progress, a considerable number of people with hearing and speech impairments still face communication challenges. Many existing tools are either incomplete or fail to be truly inclusive. This study proposes a comprehensive deep learning-based system that integrates sign language-to-text recognition with text-to-speech capabilities. Utilizing YOLO NAS and Recurrent Neural Networks (RNNs), along with techniques from natural language processing and machine learning, the system facilitates smooth, real-time communication--enhancing accessibility and social inclusion.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Communication gaps, hearing loss, speech disabilities, deep learning, sign-to-text conversion, speech-to-sign conversion, YOLO NAS, RNN, NLP, inclusivity, real-time interaction.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper presents a practical implementation framework to address the challenges in text-to-image synthesis using generative models. We propose a hybrid architecture com- binning Generative Adversarial Networks (GANs) and diffusion models to balance image fidelity, diversity, and computational efficiency. Additionally, we introduce multilingual support by leveraging pre-trained language models for cross-lingual textual understanding. Our system is evaluated on multiple datasets, demonstrating improvements in semantic accuracy, computational efficiency, and multilingual capabilities this paper presents a remote sensing-based framework for agricultural monitoring and crop yield prediction, addressing the challenges of traditional methods, which are often labor-intensive, costly, and prone to inaccuracies. By leveraging satellite imagery and advanced data analytics, the proposed system enables real-time monitoring and precise yield estimation. The integration of remote sensing technologies with machine learning algorithms, such as Random Forest Regress or and Gradient Boosting, allows for accurate modeling of the complex relationships between environmental factors and crop growth. This approach enhances decision-making in agriculture, improves data reliability, and reduces operational costs. Furthermore, the system's scalability and efficiency make it a viable solution for modern precision agriculture, promoting sustainability and trust in agricultural data.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Crop Yield Prediction, Agricultural Monitoring, Precision Agriculture, Remote Sensing, Hyper spectral Imaging, Machine Learning in Agriculture, Weather Data Analysis, Soil Analysis, Big Data in Agriculture, Geospatial Analysis, Vegetation Indices (e.g., NDVI, EVI).

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This paper presents the implementation of a healthcare chatbot powered by Retrieval-Augmented Generation (RAG), designed to provide accurate, reliable, and multilingual health assistance. The chatbot integrates natural language processing (NLP), image recognition, and speech processing technologies to offer personalized and accessible medical sup-port. It leverages open-access health databases for contextually relevant responses and includes computer vision capabilities for analyzing skin conditions. The system supports multilingual voice interactions, enhancing global accessibility to healthcare information. Our implementation demonstrates significant improvements over traditional rule-based healthcare chatbots, particularly in accuracy, multimodal interactions, and accessibility. Keywords--Healthcare, Chatbot, Retrieval-Augmented Generation, Natural Language Processing, Computer Vision, Multi-lingual Support, Artificial Intelligence.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

SymptoAI: Chatbot Powered by Retrieval-Augmented Generation (RAG)

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Floods are among the most devastating natural disasters, caus- ing loss of life, property damage, and economic disruptions. Accurate flood prediction is crucial for disaster preparedness and mitigation. This study implements machine learning algorithms, including XGBoost regression-model and K-Nearest Neighbors (KNN), combined with geospatial data to predict flood occurrences. The approach integrates hydrological, meteorological, and land-use factors to enhance prediction accuracy. The results demonstrate that machine learning models effectively analyze flood risks by identifying patterns in environmental data. The study further explores exposure assessment and land-use mapping techniques to refine predictions. The proposed system can assist authorities in proactive decision-making, minimizing flood-related damages.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Flood Prediction, Flood Detection, Street View Visualization, Google Maps, Machine Learning

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Microgrids are small-scale energy systems that may function both separately and in tandem with the larger power grid. They are made up of dispersed sources of energy, such as photovoltaics, wind, battery and traditional generators, along with advanced control systems. Although microgrids been accessible for many years, the military and college campuses were the main users until recently. Thus, while still relatively modest, the overall number of microgrids is increasing. By 2028, Guide House (formerly Navigant) predicts that the market will be close to $39.4 billion. The DC microgrid is designed to manage energy generation, storage, and distribution efficiently. A bidirectional converter is employed to facilitate seamless energy exchange between the grid and solutions for energy storage, guaranteeing the best energy utilization and storage. The simulation phase involves analyzing the microgrid's performance under varying load and generation conditions using MATLAB/Simulink.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

DC Microgrid, Boost Converter Design, Bi-directional Converter Integration, MPPT Implementation, Dynamic Load Management, Simulation and Validation

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The article discusses the solar photo voltaic-based air purifier system for automotive application. In this system, impure air is drawn through layers of pre-filters consisting of HEPA and carbon filters. To kill the germs present in the cabin air it is passed through ultraviolet lights. The system successfully filters the particulate matter of size 2.5 micrometres to 10 micrometres. The system also reduces the pungent smell present in the impure air inside the cabin. The system uses solar energy to charge the batteries independently used for solar air purifier. The solar panels are placed on the roof top of the vehicle. This makes sure that the vehicle energy source does not have the additional load to power the air purifier system. The solar energy is used for charging the batteries. The energy from the charged batteries is used for powering suction and blower pumps. The proposed system is very successful in reducing particulate matter, germs, CO2, NOX, and pungent smell from the impure air in the vehicle cabin environment. The system is environmentally friendly since it uses solar energy as a power source.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

DC Motors, ESP32, Sensors, Micro-controller, Bluetooth.

  License

Creative Commons Attribution 4.0 and The Open Definition