Volume 14 | Issue 4 |

 Abstract

The A???da?a Pur??as are not merely ancient mythological texts but living sources of philosophical wisdom relevant to modern life. This article shows how Pur??ic stories and teachings address key contemporary issues such as mental health, environmental crisis, ethical living, and existential uncertainty. Narratives from texts like the Bh?gavata Pur??a and Vi??u Pur??a symbolically represent inner struggles, helping individuals understand emotions, ego, and stress. Their ecological outlook promotes respect for nature, viewing Earth as sacred and interconnected. The path of bhakti encourages inclusivity, emotional strength, and social harmony beyond divisions of caste and gender. Philosophical ideas such as karma, m?y?, and mok?a provide meaning and direction in a materialistic and uncertain world. Overall, the Pur??as offer practical guidance for achieving personal peace, ethical responsibility, and sustainable living, making them deeply relevant even today.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

A???da?a Pur??as, Bhakti, Dharma, M?y?, Karma, Environmental Ethics, Mental Health, Sustainability, Indian Philosophy, Spirituality

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Extractive Question Answering (QA) systems have been highly developed in resource-rich languages like English mainly thanks to the existence of big annotated datasets and well-established evaluation frameworks. Nevertheless, there is still a lack of investigation on the evaluation of morphologically rich language such as Tamil. Conventional QA evaluation measure, EM , and token-level F1 score, are mostly used, which are not able to measure the boundary-level prediction behaviour especially when the answer involves the inflectional suffix or compound expressions. This paper proposes a boundary-aware evaluation framework for Tamil extractive question answering for a detailed understanding of span prediction tendencies. The proposed framework assesses the performance of several Multilingual transformer models for Tamil like, Tamil-BERT, MuRIL, XLM-RoBERTa, and IndicBERT on a Tamil QA dataset represented in a SQuAD similar format. Besides the conventional EM and F1 measures the paper reports extended evaluation measures: Strict EM Relaxed EM Span Calibration Accuracy (SCA) Character F1 Character Overlap and Average Length Deviation. Also, the errors in structure prediction are divided into expansion, truncation and wrong-region errors in order to compare the boundary prediction performances of different models. Performance differences are statistically validated through two-proportion Z-test and confidence intervals. The experimental results indicate that span calibration behavior of the multilingual transformer models differs, while they are fairly close based on EM/F1 scores.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Tamil Question Answering, Extractive Question Answering, Multilingual Transformers, Span Boundary Detection, Boundary-Aware Evaluation, Morphologically Rich Languages, Natural Language Processing.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The rapid growth of financial markets and the increasing availability of financial data have created a need for intelligent systems capable of analyzing and interpreting financial information efficiently. Traditional financial platforms often provide raw financial data but lack intelligent interpretation and decision-support capabilities. This paper presents FinanceMCP, an AI-driven financial intelligence platform designed to assist users in analyzing stocks, mutual funds, IPOs, and macroeconomic indicators within the Indian financial ecosystem. The proposed system integrates real-time financial data sources with analytical tools and artificial intelligence techniques to provide actionable insights for investors. The platform combines financial technical analysis algorithms such as Relative Strength Index (RSI) and Moving Average Convergence Divergence (MACD) with a machine learning model based on Random Forest Regression to estimate portfolio resilience during financial stress conditions. In addition, the platform integrates a Large Language Model (LLM), Claude by Anthropic, which generates natural-language explanations for financial indicators and analytical results. The architecture of FinanceMCP follows a modular design consisting of a React-based user interface, FastAPI backend services, integrated financial APIs, and a machine learning module for predictive analytics. Experimental evaluation shows that combining machine learning models with LLM-based interpretation significantly improves user understanding of financial data and supports more informed investment decisions. The proposed framework demonstrates how artificial intelligence and financial analytics can be integrated to build intelligent financial advisory platforms. Keywords: Financial Intelligence, Machine Learning, Large Language Models, Claude LLM, FinTech Analytics, Portfolio Risk Prediction, Technical Analysis

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Financial Intelligence, Machine Learning, Large Language Models, Claude LLM, FinTech Analytics, Portfolio Risk Prediction, Technical Analysis

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

To guarantee safe and effective operation, thermal power plants need constant monitoring of vital parameters including temperature, vibration, voltage, and pressure. In big industrial settings, traditional cable monitoring systems are expensive and challenging to maintain. This article describes the design and implementation of an Arduino Uno and LoRaWAN-based smart industrial monitoring system. Sensors like the MAX6675 K-type thermocouple sensor, vibration sensor, voltage sensor, and pressure sensor provide real-time data to the system. LoRaWAN is used to send the gathered data wirelessly across great distances. The suggested system offers dependable data transfer, minimal power consumption, and long-range communication (up to 2 km). The technology is appropriate for thermal power plant applications since experimental findings show precise sensing and consistent communication.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Smart Monitoring, LoRaWAN, Arduino Uno, MAX6675, Vibration Sensor.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Workforce Task Automation is an automated system which uses technological tools to automate mundane activities such as task allocation, reporting, monitoring progress, etc., for the purpose of enhancing business operations. In its findings, this study points out several flaws in the existing manual systems for managing workforces which include the delay in tasks, human errors, misunderstanding, and lack of visibility of tasks at all times. These are addressed through the proposed system which uses smart scheduling, automated alerts, role based permissions, and central data processing for increasing efficiency and collaboration. As can be seen from the above description, the use of automation enhances the accuracy of tasks, minimizes costs of doing business, and speeds up process performance

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

task automation

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

OncoEvolveAI is a self-evolving artificial intelligence framework designed to enhance tumor detection and optimize personalized therapy strategies in oncology. The system integrates advanced deep learning models with adaptive learning mechanisms to continuously improve its performance using real-time clinical data. By leveraging medical imaging, patient history, and genomic information, OncoEvolveAI provides highly accurate tumor identification and classification, reducing diagnostic errors and enabling early detection. A key feature of the framework is its ability to evolve through feedback loops, where model parameters are dynamically updated based on treatment outcomes and new data inputs. This ensures that the system remains relevant and effective across diverse patient populations and cancer types. Additionally, OncoEvolveAI incorporates predictive analytics to recommend optimized treatment plans tailored to individual patients, considering factors such as tumor characteristics, genetic markers, and response to previous therapies. The proposed framework aims to bridge the gap between diagnosis and treatment by providing an end-to-end intelligent solution for clinicians. Experimental evaluations demonstrate improved accuracy, efficiency, and adaptability compared to traditional AI models. Overall, OncoEvolveAI represents a significant step toward precision medicine, offering a scalable and robust approach to cancer care through continuous learning and intelligent decision support.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Artificial Intelligence, Breast Cancer,Deep Learning,Active Learning

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The integrity of professional and legal transactions relies heavily on authentic certificates. However, the proliferation of sophisticated digital manipulation tools has made it easy to create fraudulent documents that are indistinguishable to the human eye. Current manual verification processes are labor-intensive, slow, and prone to human error, creating significant bottlenecks. Standard automated tools like QR code verification often fail to detect visual anomalies or sub-pixel discrepancies in the document's actual content. This project proposes an automated web-based forensic tool designed to identify document tampering at a sub-pixel level. By integrating Error Level Analysis (ELA) via OpenCV to highlight JPEG compression inconsistencies and a specialized Convolutional Neural Network (CNN) for high-precision classification, the system targets a detection accuracy of over 95%. Furthermore, Explainable AI (XAI) heatmaps are generated to provide transparent visual evidence of the specific regions contributing to the forgery decision.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Artificial Intelligence, Certificate Authentication, Convolutional Neural Networks, Document Forensics, Error Level Analysis, Explainable AI.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

IoT, Smart Waste Management, Arduino UNO, Ultrasonic Sensor, ESP8266, GSM Alert, Adafruit IO, Real-Time Monitoring, Smart City, Embedded Systems.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Rapid digitalization has led to an alarming increase in electronic waste (e-waste), which poses serious environmental and health hazards. Manual segregation and disposal remain inefficient and unsafe. This paper presents WasteWise, a comprehensive software system equipped with advanced Machine Learning and Computer Vision technologies to automate e-waste detection, classification, and management. The system features multi-label waste detection using Convolutional Neural Networks (CNNs), environmental lifecycle analysis with Eco-Score computation, intelligent recycling/upcycling suggestions, a multilingual AI chatbot assistant, and a reward-based eco-wallet. Additionally, it includes a community waste heatmap for location- based waste tracking and non-waste image filtering. The model integrates a Python-based backend with a React.js frontend, delivering a scalable, interactive, and sustainable digital waste management solution.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Waste Management, Smart Waste Segregation ,Artificial Intelligence (AI), Machine Learning (ML), Deep Learning Image Classification

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Industrial effluents, Biosorbent, Antioxidant, Chromium reduction, GC-MS

  License

Creative Commons Attribution 4.0 and The Open Definition