Volume 14 | Issue 3 |

 Abstract

Brain tumor diagnosis using Magnetic Resonance Imaging (MRI) remains a challenging task due to tumor heterogeneity, complex morphology, and variability across imaging modalities. Recent advances in deep learning, particularly Convolutional Neural Networks (CNNs), have significantly improved automated detection and classification performance. However, conventional CNNs often struggle to capture global contextual dependencies and long-range spatial relationships. To address these limitations, hybrid architectures integrating attention mechanisms--such as channel, spatial, self-attention, and Transformer-based modules--have emerged as state-of-the-art solutions. This paper presents a comprehensive review of hybrid CNN-attention frameworks for brain tumor detection and multi-class classification. The review systematically analyzes architectural designs, datasets, preprocessing strategies, evaluation metrics, and performance trends. A taxonomy of hybrid models is proposed, including CNN + attention modules, CNN + Transformer hybrids, and multi-branch fusion networks. Comparative analysis highlights that hybrid attention-based models consistently outperform standalone CNNs, achieving accuracy levels exceeding 98% on benchmark MRI datasets. Furthermore, this review identifies critical research gaps, including lack of clinical generalization, dataset bias, interpretability limitations, and computational overhead. Future directions focusing on explainable AI, federated learning, and lightweight deployment are discussed. The study aims to provide a structured foundation for researchers developing next-generation intelligent diagnostic systems.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Brain Tumor Classification, MRI, Hybrid CNN, Attention Mechanisms, Vision Transformers, Deep Learning, Medical Imaging

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

This study looks at the HR policies and practices of small-scale businesses, with a focus on how these organizations manage their workforces to promote organizational expansion and enduringness. Small firms often suffer with a lack of resources, financial constraints, and the need for effective HR procedures in order to remain competitive in the market. This research looks at key HR practices, including recruiting, training, performance evaluations, compensation, and employee relations, and assesses how they impact employee satisfaction, productivity, and overall corporate success. Using qualitative and quantitative data collected from a range of small-scale industries, the study illustrates the unique HR practices employed by small enterprises and how they adapt their strategies to match the specific requirements of their operations.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Human Resource Management, Small-Scale Manufacturing Industry, HR Policies, Recruitment, Training and Development, Performance Appraisal, Employee Satisfaction, Workforce Productivity, Organizational Growth.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Abstract The Telangana Movement (1969-1975) represents a crucial phase in post-independence India's regional political history. While the agitation for a separate Telangana state has been widely studied, the role of women participants remains underexplored. This paper examines the contributions of women activists, leaders, students, and grassroots participants in the movement. It argues that women were not passive supporters but active agents who shaped the social base, ideological framework, and cultural expression of the agitation. Despite their significant involvement, patriarchal structures and historiographical biases marginalized their contributions. This study attempts to reconstruct women's participation and situate it within broader debates on gender, regional identity, and political mobilization.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Keywords Telangana Movement, Women Activists, Sadalakshmi, Eswari Bai, Gender History, Regional Politics, Social Movements, 1969 Agitation

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Type 2 Diabetes Mellitus (T2DM) is a major global health concern associated with multiple complications, among which Diabetic Peripheral Neuropathy (DPN) is highly prevalent and disabling. DPN leads to impaired balance, reduced lower extremity muscle strength, and increased anxiety levels, thereby increasing fall risk and reducing quality of life. Exercise therapy is widely recommended as a non-pharmacological intervention to improve both physical and psychological outcomes in individuals with DPN. This randomized clinical trial aimed to compare the effects of Interval Walking Training (IWT) and Nordic Walking (NW) on balance, lower limb muscle strength, and anxiety levels in individuals with DPN. A total of 66 participants aged 45-60 years were randomly allocated into two groups: Interval Walking Training (n = 33) and Nordic Walking (n = 33). Both groups underwent supervised training sessions for a specified duration. Outcome measures included the Berg Balance Scale (BBS) for balance, Manual Muscle Testing (MMT) for lower extremity strength, and the Generalized Anxiety Disorder-7 (GAD-7) scale for anxiety. Statistical analysis was performed using paired t-tests for within-group comparisons and independent t-tests for between-group comparisons. The results demonstrated that both groups showed statistically significant improvements in balance, muscle strength, and anxiety levels following the intervention (p < 0.05). Within-group analysis revealed significant improvements in BBS, MMT, and GAD-7 scores in both groups. However, between-group comparison demonstrated no significant difference in balance (p = 0.945) and anxiety levels (p = 0.531), while Interval Walking Training showed significantly greater improvement in lower extremity muscle strength compared to Nordic Walking (p < 0.05). In conclusion, both Interval Walking Training and Nordic Walking are effective in improving balance and reducing anxiety in individuals with Diabetic Peripheral Neuropathy; however, Interval Walking Training is superior in enhancing lower extremity muscle strength, supporting its inclusion in rehabilitation protocols for individuals with DPN.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Type 2 Diabetes Mellitus, Diabetic Peripheral Neuropathy, Interval Walking Training, Nordic Walking, Balance, Muscle Strength, Anxiety.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

In modern educational institutions, secure and efficient student entry management has become a significant challenge due to increasing security concerns and large student populations. Traditional methods such as manual attendance registers and physical identity card verification are inefficient, error-prone, and vulnerable to misuse. These systems lack real-time validation, centralized monitoring, and secure digital record maintenance. This research proposes a QR Code Based Secure Student Entry Management System that integrates encrypted QR authentication with real-time server validation and automated digital logging. Each registered student is assigned a unique, time-sensitive QR code generated using secure tokenization techniques. The QR code is scanned at entry checkpoints, and the system verifies its authenticity through a centralized database server before granting access. The system enhances operational efficiency, prevents identity duplication, ensures secure communication using encryption protocols, and maintains tamper-proof digital logs. Experimental evaluation demonstrates significant improvement in processing speed, accuracy, and administrative monitoring compared to traditional entry systems. The proposed solution provides a scalable, cost-effective, and secure framework suitable for academic institutions.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

QR Code Security, Access Control, Real-Time Validation, Digital Logging, Student Authentication, Secure Entry System

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Cold chain management plays a crucial role in maintaining the quality and safety of temperature-sensitive products such as vaccines and perishable food items. Improper temperature control during storage and transportation can lead to vaccine degradation and food spoilage, resulting in serious health and economic consequences. This survey focuses on IoT-based cold chain monitoring systems that use sensors, microcontrollers, and communication technologies to ensure continuous temperature and location tracking. Existing solutions using DHT22 temperature and humidity sensors, u-blox NEO-M8N GPS module, Wi-Fi communication, and cloud platforms are studied and analyzed. The survey highlights system capabilities, limitations, and implementation challenges. The paper aims to provide a clear understanding of current cold chain monitoring approaches and supports the development of an efficient, low-cost, and reliable IoT-based monitoring system.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Cold Chain Monitoring, IoT, Vaccine Transportation, Food Logistics, Temperature Sensors, u-blox NEO-M8N GPS Module, Cloud Monitoring

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The foundational decision regarding software architecture is a critical determinant of a system's long-term success, directly influencing factors such as scalability, maintainability, and operational complexity. The pervasive shift toward distributed computing has fueled a contemporary debate regarding the suitability of traditional monolithic structures versus the emergent microservices paradigm. This research provides a comparative analysis addressing the architectural dilemma faced by organizations when designing modern, resilient systems. We investigate both architectures across key dimensions, including scalability, deployment agility, fault isolation, and organizational alignment, using a structured analytical methodology grounded in established distributed systems theory. High-level findings indicate that while monolithic architectures offer simplicity in initial development and unified data management, they inherently introduce coupling that impedes independent scaling and rapid deployment. Conversely, microservices facilitate superior decoupling, enabling independent deployment and specialized scaling, yet introduce significant complexity concerning operational overhead, inter-service communication, and distributed data management. The analysis concludes that the optimal architectural choice is highly contextual, relying on a systematic evaluation of specific business requirements, organizational maturity, and system complexity rather than the adoption of a universally superior model

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Monolithic Architecture, Microservices, Software Scalability, DevOps, Fault Isolation, CAP Theorem, Conway's Law, Distributed Systems, Deployment Agility, System Design.

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

The The need for intelligent systems that can extract pertinent information from many documents has increased due to the increasing growth of digital textual data. Conventional transformer-based language models perform well in natural language comprehension, but they are constrained by hallucination problems and context window limitations. A Retrieval-Augmented Transformer framework for question- answering and multi-document summarization is proposed in this study. Document chunking, transformer-based response generation, similarity-based retrieval utilizing FAISS, and dense vector embeddings are all integrated into the system. The model increases scalability and factual grounding by only accessing pertinent contextual information prior to generation. Compared to standalone transformer models, experimental findings show increased reaction accuracy and less creation of extraneous material. By retrieving only relevant contextual information before generation, the model improves factual grounding and scalability. Experimental observations indicate improved response accuracy

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Retrieval-Augmented Generation(RAG), FAISS,Multi-Document Summarisation, Large language model, Multi-format Document Processing

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Background: India's judicial system manages over 50 million pending cases with average disposition times exceeding 15 years, creating severe delays in justice delivery. Artificial Intelligence and Machine Learning offer potential solutions through automated legal document analysis and case outcome prediction. Objectives: This research investigates Retrieval-Augmented Generation (RAG) combined with Large Language Models (LLMs) for supporting legal document analysis, case outcome prediction, and explainable reasoning in the Indian judicial context. Six research questions guide the investigation: (RQ1) effectiveness of document retrieval methods, (RQ2) answer accuracy and factual grounding, (RQ3) pipeline synergy, (RQ4) user trust and explainability impact, (RQ5) multimodal interface accessibility, and (RQ6) responsible deployment safeguards. Methodology: A three-tier evaluation framework was employed: (i) retrieval tier evaluating hybrid (semantic + keyword) retrieval on 600 Indian court judgment documents, (ii) generation tier assessing LLM output quality on 30 test questions with 5 expert evaluators, and (iii) user-centered tier with 15 legal professionals (lawyers, judges, legal researchers) using System Usability Scale and user trust questionnaires. Results: The RAG system achieved 83.3% exact match accuracy (vs. 71% for LLM-only baseline), with hallucination rate of 5.1% (vs. 14.2% baseline, 64% reduction). Citation accuracy reached 94.7%, indicating strong source grounding. User trust improved by 41% (2.9 -> 4.1 on 5-point scale) with source transparency. Retrieval precision was 0.72 (Precision@5) and 0.84 (Recall@10). System Usability Score: 76.3 (Good category). Speech-to-text achieved 88.3% overall accuracy but showed accent bias (83.5% Indian English vs. 97% standard English). Performance varied by case type (Criminal: 85.2%, Appeals: 80.3%). Deployment Readiness: Core functionality is production-ready. Critical challenges include hallucination mitigation (target: <2%), Indian English ASR fine-tuning (target: 92%+), bias detection system development (current: 62.5%, target: 85%+), and regulatory framework establishment. Conclusions: RAG systems achieve superior accuracy and factual grounding compared to LLM-only approaches. Source transparency dramatically increases user trust, validating explainability-by-design approaches. Legal AI deployment requires domain-specific methods, careful bias auditing, clear regulatory frameworks, and stakeholder collaboration. The research provides evidence-based guidance for AI-assisted legal research systems and contributes to responsible AI deployment frameworks for high-stakes domains.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Retrieval-Augmented Generation, Explainable Artificial Intelligence, Legal AI, Natural Language Processing, Machine Learning, Indian Judiciary, Case Outcome Prediction, Responsible AI, Multimodal Interfaces Conflict of Interest

  License

Creative Commons Attribution 4.0 and The Open Definition

 Abstract

Shirpur, a modest town in Maharashtra's Dhule district, is gradually emerging as a promising candidate for tourism development, owing to its integrated strengths in water management, agro-industrial prominence, educational ecosystem, and cultural-spiritual heritage. This research aims to assess Shirpur's untapped potential through a critical review of secondary data from official reports, case studies, and academic literature. Key variables explored include tourism infrastructure, environmental sustainability, regional economy, and public-private participation. The findings suggest that Shirpur, with the right strategic investment and branding, could evolve into a model for sustainable, rural-industrial tourism in Maharashtra.

Licence: creative commons attribution 4.0

  License

Creative Commons Attribution 4.0 and The Open Definition

 Keywords

Shirpur, rural tourism, eco-tourism, sustainable development, industrial visits, Maharashtra

  License

Creative Commons Attribution 4.0 and The Open Definition