Volume 12 | Issue 5 |

 Abstract

The proliferation of digital documents, particularly in PDF format, necessitates innovative solutions for efficient information retrieval and interaction. In response to this growing demand, we introduce a pioneering framework enabling users to engage in natural language conversations with PDF documents through a chatbot interface powered by Large Language Models (LLMs). Our system addresses the inherent challenges of document interaction, including complex content structures, diverse document types, and varying user needs. By harnessing the capabilities of LLMs, users can upload PDF documents and seamlessly converse with the chatbot to extract specific information, pose queries, and navigate through the document's contents. The development of our chatbot framework is motivated by the imperative to enhance accessibility and usability in document management systems across diverse domains such as academia, research, and corporate environments. Through comprehensive document preprocessing techniques and sophisticated LLM-based conversational agents, our framework empowers users with intuitive and efficient means to interact with PDF documents. We underscore the significance of our approach through empirical evaluations and real-world case studies, demonstrating its effectiveness in handling multifaceted document structures and providing accurate responses to user inquiries. This research represents a significant step forward in augmenting human-computer interaction paradigms, offering a user-centric approach to unlocking the wealth of knowledge embedded within PDF documents.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Chatbot, PDF Interaction, Large Language Models, Document Understanding, Natural Language Processing, Information Retrieval.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The Fall Detection System is a critical component in ensuring the safety and well-being of individuals, particularly the elderly population. Falls are a leading cause of injury and mortality among seniors. This report presents a comprehensive study of a Fall Detection System designed to promptly and accurately identify instances of falls, providing immediate alerts and assistance. The system incorporates a range of sensors, data processing techniques, and real-time communication mechanisms to achieve its objectives. The primary objectives of this research are to design an efficient and reliable Fall Detection System, evaluate its performance in simulated scenarios, and assess its practicality in real-world applications. Various sensor technologies, including accelerometers, gyroscopes, and pressure sensors, are employed to detect abrupt changes in motion and body position. Data from these sensors are processed through deep learning algorithms to distinguish between normal activities and fall events.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

ML,DL

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Accurately identifying pharmaceuticals, such as pills, tablets, and capsules, is essential for ensuring patient safety, enhancing medication adherence, and streamlining healthcare delivery. Traditional methods rely on human judgment and manual processes and are susceptible to errors that can lead to adverse patient outcomes. This paper investigates the utilization of machine learning (ML), deep learning (DL), and hybrid algorithms to improve the precision and reliability of pill identification. We explore various ML techniques, including Support Vector Machines (SVM) and Random Forests, alongside DL methods such as Convolutional Neural Networks (CNNs), which are particularly adept at image recognition tasks. By combining these approaches in a hybrid model, we perform better in identifying and classifying pharmaceuticals based on their physical characteristics. Our experimental results demonstrate that these advanced techniques can significantly mitigate medication errors, enhance operational efficiency, and provide a robust framework for scalable pharmaceutical identification systems. This study highlights the transformative potential of ML and DL in advancing healthcare safety and efficacy.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Pharmaceutical Identification, Pill Detection, Machine Learning, Deep Learning, Convolutional Neural Network (CNNs), Support Vector Machines (SVM), Random Forests, Hybrid Algorithms, Medication Safety, Healthcare Technology, Image Recognition, Patient Safety.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

In an era where global markets are intricately connected, the proliferation of counterfeit products has become a significant challenge, undermining trust and violating consumer agreements. Traditional methods to enforce trust and agreements are often circumvented, presenting a loophole that is exploited frequently. Blockchain technology offers a robust solution to these challenges through its inherent features of transparency, immutability, and decentralization. This paper discusses the problem of counterfeit products that travel through extensive supply chains, often altering the original agreement between producer and consumer. By implementing blockchain technology, this study explores a system that not only tracks the journey of products through the supply chain but also provides consumers with reliable means to verify product legitimacy. Our analysis begins by identifying the scale of the counterfeiting problem, followed by an assessment of current blockchain implementations in various industries. We evaluate the effectiveness of these technologies in maintaining the integrity of supply chains and restoring consumer trust. The paper concludes with strategic recommendations for integrating blockchain technology to secure supply chains against counterfeiting threats. This approach promises to significantly reduce the incidence of counterfeit goods and enhance consumer confidence by ensuring product authenticity from origin to end- user.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Blockchain, Counterfeit Products, Supply Chain Integrity, Consumer Trust, Product Authenticity, Transparency, Decentralization.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This study pioneers the development of an advanced digital tool utilizing artificial intelligence (AI) for the classification and informational enrichment of a wide array of food items, transcending the initial focus on fruits and vegetables. Aimed at revolutionizing the way individuals' access and interact with nutritional information, our project leverages a sophisticated convolutional neural network (CNN) alongside natural language processing (NLP) technologies. These are designed not only to accurately identify diverse food items from images but also to generate detailed, user-friendly content regarding their nutritional content, health benefits, and culinary uses. The research embarked with a comprehensive dataset, initially comprising over 10,000 images of 500 different fruits and vegetables, achieving an impressive 98% accuracy in image classification. This success laid the groundwork for the subsequent expansion of our dataset to include a broader spectrum of food items, addressing the critical need for a more inclusive and versatile nutritional information tool. Our findings, underscored by a pilot user study with 50 participants, indicate significant engagement and positive feedback, suggesting the tool's potential to significantly impact public health by enhancing nutritional awareness and food literacy across diverse populations. Moreover, the planned expansion to include a wider variety of food items reflects our commitment to adapt and respond to the evolving dietary needs and preferences of the global community. This research not only demonstrates the feasibility of applying AI in the domain of nutritional education and public health but also highlights the potential for such technologies to foster informed food choices, support dietary planning, and contribute to the broader goals of improving public health and nutritional awareness on a global scale.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Artificial Intelligence, Food Classification, Nutritional Information, Dietary Education, Public Health, Convolutional Neural Network, Natural Language Processing

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The application of artificial intelligence (AI) techniques to information technology (IT) industry human resource (HR) operations is the subject of this study. It uses empirical data from surveys and interviews with IT stakeholders and HR professionals, along with a thorough analysis of the body of literature already in existence, to investigate the adoption, application, and effects of AI tools on a range of HR functions, such as hiring, performance management, and employee engagement. According to the research, AI is transforming HR procedures by facilitating data-driven decision-making, automating processes, and improving employee experiences. But issues have been raised about algorithmic bias, data privacy, and insufficient knowledge.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Artificial intelligence (AI), human resource (HR) operations, the information technology (IT) sector, Workforce analytics, automation, data-driven decision-making, employee experience, challenges, and data privacy.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Early disease detection is paramount for effective healthcare management. In this research, titled "MediForecast: Multiple Disease Prediction," we address this critical challenge by harnessing the power of machine learning, specifically employing the Support Vector Machine (SVM) classifier algorithm. Focusing on heart disease, Parkinson's disease, and diabetes, we explore innovative approaches to predict these conditions accurately. Our methodology involves meticulous data collection, preprocessing, and feature selection tailored for each disease. We employ the SVM classifier to create robust prediction models. Our implementation demonstrates the practical application of these models, showcasing their effectiveness in diagnosing the aforementioned diseases. The results reveal promising outcomes, indicating high accuracy, sensitivity, and specificity in disease prediction. By empowering medical professionals with timely and precise predictive capabilities, our research contributes significantly to the advancement of healthcare practices. We highlight the transformative potential of machine learning, particularly the SVM classifier, in revolutionizing disease diagnosis, paving the way for a healthier future.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Diabetes, Heart, Parkinsons, SVM Classifier

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The Metamal-D-Alert project is an advanced malware detection system designed to address the rising threats of malware, which often lead to severe data breaches, financial losses, and operational disruptions. This initiative employs sophisticated machine learning techniques, specifically Convolutional Neural Networks (CNNs) and Bidirectional Gated Recurrent Units (BiGRU), to identify abnormal patterns indicative of malware. Using the custom MetaMal-D-Alert dataset for Windows environments, the system effectively detects both known and zero-day malware attacks. Implemented as a scalable web application with Docker containers, MetaMal-D-Alert ensures seamless integration and deployment across various platforms. Enhanced by API analysis and Natural Language Processing (NLP), it offers comprehensive dynamic and hybrid threat analysis. Experimental results show that Metamal-D-Alert significantly outperforms traditional signature-based methods, providing superior accuracy and efficiency in identifying malicious activities.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Malware detection, Neural Networks, API calls, dynamic analysis, Hybrid analysis API, Cybersecurity, Machine Learning, NLP, Deep Learning.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

In regions prone to seismic activity like the Pacific Ring of Fire, ensuring the structural integrity of buildings is paramount to mitigating the devastating impact of earthquakes [1]. Retrofitting stands as a crucial process aimed at fortifying existing structures to withstand seismic forces [2]. Among various retrofitting devices, dampers play a pivotal role in absorbing the kinetic energy generated during seismic events [3]. Dampers counteract the tensile and compressive forces that jeopardize structural integrity, with their arrangement contingent upon materials and principles employed [4].Strategic placement of dampers is imperative to maximize their effectiveness in absorbing movement and forces [5]. They are typically positioned diagonally across floors or connected to opposing corner sides, depending on the damper type, enhancing the building's resilience [6]. Seismic retrofitting, including the integration of girders, emerges as quintessential to modify existing structures and render them more resilient against seismic activity [7].Recent scientific focus has shifted towards retrofitting methods to bridge the gap in seismic safety for existing buildings [8]. Various techniques such as RC/mortar jacketing, steel jacketing, and FRP jacketing have been explored to enhance flexural and shear capacities [9]. FRP jacketing, particularly, stands out for its ease of installation and competitive effectiveness [10]. Additionally, innovative approaches like TRM jacketing have shown promise in increasing strength and deformation capacity [11].Empirical research, such as that conducted by Saeedi and Abbasi (2017), highlights the efficacy of retrofitting techniques like girder-column connections in enhancing the seismic resilience of aging reinforced concrete buildings [12]. Experimental studies, like those by Almeida et al. (2016), underscore the effectiveness of retrofitting solutions such as BRBs in limiting structural damage and improving resilience [13]. Furthermore, advancements in analytical methods and numerical simulations contribute to refining retrofitting strategies and understanding structural behavior under seismic loading conditions [14].Integrating girders as a retrofitting solution offers a promising avenue for enhancing the seismic resilience of aging reinforced concrete buildings [15]. Through comprehensive investigations and innovative approaches, researchers have demonstrated significant improvements in structural integrity and seismic performance. This holistic approach not only addresses immediate safety concerns but also contributes to long-term sustainability and resilience against earthquakes. As seismic threats persist, proactive retrofitting measures remain crucial in safeguarding lives and infrastructure.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Seismic retrofitting, earthquake engineering, structural resilience, PET FRP, metal dampers, historic buildings, GESB system, cross-section dampers, external sub-structure retrofitting, friction dampers, shaking table tests.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Abstract- With the world becoming more interconnected domestic abuse stands as a pervasive issue transcending geographical and social boundaries, impacting individuals across the globe. This research paper delves into the multifaceted aspects of domestic violence, with a specific focus on the regulatory actions and initiatives implemented by the governments of India and Ethiopia. It conducts a comparative assessment of the scope and effectiveness of domestic abuse laws in these two nations, drawing attention to both the differences and commonalities in their approaches. India has established a panoramic legal framework, exemplified by the Protection of Women from DVA, while Ethiopia has similarly introduced legislation such as the Criminal Code of the Federal Democratic Republic of Ethiopia. This paper rigorously evaluates the impact of these regulations in both countries and analysis the progress made in addressing domestic abuse, including support services, committee reports and surveys etc. By comparing these legal frameworks and practical measures, this research contributes valuable insights to the global fight against domestic violence, deepening our understanding and paving the way for more effective strategies to protect individuals from domestic abuse on a global scale.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Domestic abuse, domestic violence, legal frameworks, India, Ethiopia, government initiatives, comparative assessment, Protection of Women from Domestic Violence Act (DVA), Criminal Code of Ethiopia, regulatory actions, support services, committee reports, surveys, legislative impact, global interconnectedness, social boundaries, pervasive issue, effectiveness of laws, addressing domestic abuse, international strategies

  License

Creative Commons Attribution 4.0 and The Open Definition