Authors: Ujjwal Kumar, Ritik Kumar, Kavita kumari, Nitika Vats

Abstract: Apigenin is a low-toxicity flavonoid with several beneficial bioactivities. It is a secondary metabolite bioflavonoid that shows pharmacological activities such as antibacterial, anticancer, antidiarrheal, antiemetic, and hemostatic effects. Apigenin is extracted from different parts of parsley plants by using the Soxhlet Extraction method. Parsley (Petroselinum crispum), a flowering species of the Apiaceae family, is native to Greece, Morocco, and the former Yugoslavia. In traditional medicine, parsley has been used as a carminative, gastrotonic, diuretic, urinary tract antiseptic, anti-urolithiasis, antidote, and anti-inflammatory agent. It is also used for treating dermal diseases, amenorrhea, dysmenorrhea, gastrointestinal disorders, hypertension, cardiac and urinary diseases, otitis, cold, and diabetes. Apigenin is a phytochemical that occurs along with tannins, alkaloids, triterpenoids, steroids, flavonoids, and saponins. Rich in parsley, celery, celeriac, and chamomile tea, apigenin has health-promoting properties, including use as a natural sleep aid, antidiabetic, and anticancer compound. This flavonoid induces relaxation as it binds to brain receptors that promote sleep. The main aim of this paper is to extract apigenin flavonoid from parsley plant parts. Parsley leaves and flowers are abundant in phenolic compounds, present as aglycones (flavones and flavonols) and glycosides. In this study, apigenin was extracted from parsley leaves using Soxhlet extraction, followed by hydrolysis and recrystallization. A combination of apigenin and lecithin was also synthesized using a solvent method. Several extraction parameters were tested to evaluate yield, with Soxhlet extraction 5.5 h, 65 °C, solid-to-solvent ratio 8:400 as the reference and the purification done by colume chromotography. UV-Visible analysis confirmed that the structure of apigenin remained stable after extraction and purification.

DOI: https://doi.org/10.5281/zenodo.18058776

 

Published by: vikaspatanker

Authors: Dr. Rachna Rana, Er. Gundeep Kaur, Er. Manpreet Kaur, Mr. Sachin Sharma

Abstract: Artificial Intelligence (AI) is reshaping educational systems worldwide through personalized learning, predictive analytics, intelligent tutoring systems, automation, and institutional decision-support technologies. AI applications in education have transitioned from experimental prototypes to widely adopted tools used for assessment, student support, curriculum design, and governance. This paper presents a comprehensive analysis of the current landscape of AI in education, with emphasis on machine learning (ML) frameworks, learning analytics (LA), natural language processing (NLP), and predictive analytics used for monitoring academic quality assurance (QA). The paper synthesizes findings from recent empirical and conceptual studies, discusses the system-level implications of AI-enabled educational data mining, and identifies ethical, pedagogical, and institutional challenges that influence adoption. A section is dedicated to the integration of AI-driven predictive models into QA processes, including early warning systems, risk-prediction algorithms, and data-driven continuous-improvement frameworks. The paper concludes with recommendations for responsible AI deployment, future research trajectories, and policy considerations.

DOI: http://doi.org/10.5281/zenodo.18081433

Published by: vikaspatanker

Authors: Vaibhav Sinha, Abhishek Kumar Singh, Dr. Partap Singh

Abstract: The integration of 5G technology and Artificial Intelligence (AI) marks a transformative phase in digital communications and intelligent connectivity. As 5G networks offer unprecedented speed, ultra-low latency, and massive device connectivity, AI brings the intelligence required to optimize and automate 5G systems. This research paper critically examines how AI empowers 5G networks, explores key applications, discusses challenges, and highlights future prospects across industries. With supporting pictorial references, the paper presents a comprehensive, humanized view suitable for academic and professional audiences.

DOI: http://doi.org/10.5281/zenodo.18053277

Published by: vikaspatanker

Authors: Satyaki Adak

Abstract: Distributed Machine Learning (DML) enables efficient training over massive datasets by distributing computation across multiple nodes; however, it also increases vulnerability to data poisoning attacks, where adversaries inject malicious or mislabeled data to corrupt the learning process. Ensuring model integrity in such environments is a critical security challenge. This project classifies DML systems into basic-DML and semi-DML based on whether the central server participates in dataset training. For the basic-DML scenario, a novel cross-learning–based data poisoning detection scheme is proposed, where training results from distributed workers are compared through multiple training loops to identify anomalous behaviour. A mathematical model is developed to determine the optimal number of training loops that maximizes detection accuracy while minimizing overhead. For the semi-DML scenario, an enhanced poison detection mechanism is introduced by leveraging the central server’s computing resources, along with an optimal resource allocation strategy to reduce unnecessary computation. Experimental results demonstrate that the proposed schemes significantly improve model accuracy—up to 20% for Support Vector Machines and 60% for Logistic Regression in basic-DML—while reducing wasted resources by 20–100% in semi-DML. The proposed framework offers a general, efficient, and scalable defence against data poisoning attacks in distributed learning environments.

DOI: http://doi.org/10.5281/zenodo.18051593

Published by: vikaspatanker

Authors: Anees Ahmed Pinjari, Prashant Yelmar

Abstract: Wildfires have become one of the greatest and the most ongoing environmental hazards in the state of California, with a profound ecological loss, finances, and loss of life. Spatio-temporal dynamics of wildfire incidences are of great importance to the successful detection of the threat, mitigation planning, and allocation of resources. This paper is a Spatio-analytical analysis of wildfire threat in California based on incident-level data between 2013 and 2019. The analysis will incorporate time trends, spatial dispersion, fire intensity, duration, loss of life, and fire management efforts to recognize at risk areas and the changing nature of wildfires. Findings indicate that there was a strong increase in the severity and duration of wildfires in 20172018, with an excessively high proportion of acreage and deaths being agglomerated around a limited number of large events. Spatial analysis points to the areas of constant hotspots of wildfires in southeastern California, where the presence of fires correlates closely with population density and administrative fire management areas. The results also show that the efficiency of wildfire response increases after a severe fire season, as evidenced by diminished person deployment compared to the severity of the incidence in the following years. Revealing the essential spatial trends and temporal changes in the behaviour of wildfires, this investigation provides practical information to detect threats in time, mitigate them, and use the time as a policy to prevent wildfires. The suggested analytical framework is a data-based source of the improvement of wildfire preparedness and assisting in predictive and decision-support systems in the future.

Published by: vikaspatanker

Authors: Umar Manzoor Shah

Abstract: This study examined how four newspapers based in Kashmir portrayed the Ladakh violence in response to the region's demand for inclusion in the 6th Schedule of the Indian Constitution and the conferment of statehood. The conflict commenced on September 24, 2025, between local protesters and law enforcement in Leh, Ladakh. The Buddhists and Muslims in this region have collaboratively established an organisation advocating for Ladakh's elevation from a Union Territory to a full state, as well as the implementation of the Sixth Schedule of the Indian Constitution to safeguard their environment, land, and employment opportunities. Discussions between the Government of India and local leadership have persisted for several years; however, these negotiations reached an impasse on September 24 due to violence. Four citizens were fatally shot via police gunfire, and over 100 sustained injuries as the crowd escalated into violence during a protest demonstration in Leh's major market. The regime instituted a curfew and arrested environmentalist Sonam Wanchuk, a prominent advocate for the cause. A content analysis of four newspapers based in Kashmir was done to ascertain the overall pattern of coverage and the degree and existence of framing regarding this subject. The analysis encompassed the frame utilised, tonal variations and article count regarding the situation in Ladakh. One hundred seventy newspaper articles were extracted from archives and examined from September 1, 2025, to October 5, 2025. The study revealed that law and order frames were utilised more frequently than political and human frameworks. The coverage in regional newspapers of Kashmir was predominantly pro-government. The findings indicate a significant application of law and order, as well as administrative frameworks, in the reporting of violence and its aftermath.

DOI: http://doi.org/10.5281/zenodo.18045860

Published by: vikaspatanker

Authors: Vipul Kanhere, Suraj Sonar, Atharva Awale, Pranav Shinde, Savita Biradar

Abstract: Software developers dedicate a substantial portion of their time to comprehending existing code, a challenge that intensifies as codebases grow in scale and complexity. Code Explanation Generators and Code Insight SaaS platforms have emerged as promising solutions, leveraging large language models to transform source code into accessible natural language explanations. This survey presents a comprehensive examination of code explanation technologies, tracing their evolution from traditional template-based and rule-based approaches through neural sequence models to contemporary LLM-powered systems. We establish a taxonomic framework for categorizing explanation tools across dimensions including target audience, explanation granularity, architectural approach, and deployment model. Our analysis encompasses commercial platforms, open-source implementations, IDE integrations, and lightweight web applications built on frameworks such as Streamlit that enable rapid development and free cloud deployment. The comparative analysis reveals significant consolidation around large language model approaches, with differentiation increasingly based on interface design, prompting strategies, and deployment architectures rather than fundamental algorithmic differences. Despite remarkable progress in explanation quality and accessibility, we identify persistent gaps including primitive granularity adaptation mechanisms, absent interpretability features for reliability assessment, inadequate privacy-preserving deployment options, limited contextual awareness beyond isolated code snippets, and evaluation methodologies that fail to capture developer-centric comprehension outcomes. Based on these findings, we propose future research directions encompassing improved evaluation frameworks grounded in task- based assessment, interpretable explanation generation with confidence indication, domain-specific adaptation for specialized contexts, and responsible deployment practices addressing privacy, accuracy, and equitable access. This survey provides structured guidance for researchers advancing code explanation capabilities and practitioners developing or adopting explanation tools.

Published by: vikaspatanker

Authors: Rishi Kumar

Abstract: Timely detection of plant stress is a major challenge in agriculture, as conventional monitoring methods rely on visible symptoms that often appear after irreversible physiological damage has occurred. This delay contributes to significant crop losses, inefficient water use, and reduced agricultural sustainability. PhytoLink is a conceptual plant-monitoring system designed to address this limitation by translating plants’ internal electrical signals into early, actionable warnings. Plants generate distinct bioelectrical responses to stress factors such as water deficiency, disease onset, and environmental changes, which can be detected before external symptoms become visible. PhytoLink proposes a non-invasive bio-electronic interface that captures these electrical signals, processes them using signal analysis techniques, and delivers clear alerts to farmers and gardeners, enabling intervention more than 48 hours in advance. By shifting plant care from reactive to proactive management, PhytoLink has the potential to reduce crop losses by 30–50%, conserve water resources, and improve decision-making in precision agriculture. This paper presents the conceptual framework, working principle, applications, and future scope of PhytoLink as an innovative tool for sustainable and intelligent plant care.

DOI: http://doi.org/10.5281/zenodo.18045548

Published by: vikaspatanker

Authors: Nimesh Agrawal, Mrs. Priyanka Bamne, Mr. Ranjeet Vishwakarma

Abstract: The exponential growth of e-commerce has led to an overwhelming abundance of products, making it challenging for consumers to find items that align with their preferences. Product recommendation systems have emerged as essential tools to enhance user experience by providing personalized suggestions. This paper delves into various recommendation methodologies, including collaborative filtering, content-based filtering, hybrid approaches, and deep learning techniques. It also explores the challenges faced in implementing these systems, such as scalability, cold-start problems, and data sparsity. Furthermore, the paper discusses evaluation metrics and real-world applications, providing insights into the effectiveness of different recommendation strategies.

DOI: http://doi.org/10.5281/zenodo.18046514

Published by: vikaspatanker

Authors: Dinesh Kunwar Patel

Abstract: This paper describes the design, software architecture, capabilities, and educational deployment of Robot Shalu, a low-cost, social and educational humanoid robot developed by a schoolteacher using largely recycled materials. Shalu demonstrates multilingual natural-language interaction (reported 47 languages), basic perception and memory, scripted and AI-assisted pedagogical interactions, and low-cost hardware solutions intended for real-world classroom integration in resource-constrained settings. We present the engineering choices, discuss human–robot interaction (HRI) considerations, review public reception and recognition, identify limitations, and propose rigorous evaluation protocols and next-step research to validate Shalu for broader academic acceptance. The paper aims to bridge maker-community innovation and formal scientific evaluation to support adoption of affordable humanoid educational agents.

Published by: vikaspatanker