Daily Archives: June 17, 2026

Authors: Professor Sonali Dongare, Priyanshu Singh, Aditya Amup

Abstract: Motor impairments such as ALS, locked-in syndrome, and cerebral palsy severely limit an individual's ability to interact with digital systems using conventional input devices. This paper presents GazeSpeak, an AI-powered Eye Gaze Communication System that enables motor-impaired users to communicate through voluntary eye movements alone. The system extracts real-time gaze coordinates using OpenCV and MediaPipe, maps them onto interactive screen elements via a TensorFlow regression model, and integrates a transformer based NLP module for context-aware word prediction. A dwell-based selection mechanism activates interface targets without any physical input. Experimental evaluation across twenty participants demonstrates a gaze detection accuracy of 94.2%, end-to-end latency of 38ms, top-3 word prediction accuracy of 87.6%, and communication throughput of 10.6 WPM, with a System Usability Scale score of 84.4 confirming excellent user acceptance. The results establish GazeSpeak as an effective, open-source, and cost-accessible assistive communication platform for real-world deployment.

DOI: https://doi.org/10.5281/zenodo.20735406

Published by: vikaspatanker

Authors: Aarush Kukade, Advait Deogade, Atharva Mane, Dr. Saurabh Saoji

Abstract: In the digital banking era, effective marketing lead generation depends on leveraging hetero-geneous, multimodal customer data. Traditional predictive models primarily rely on tabular attributes, overlooking the relational and contextual information inherent in customer networks. This paper proposes a Graph Neural Network (GNN)-based framework that integrates multi-modal data—including structured CRM attributes, transactional records, and unstructured call transcript text—to predict customer lead conversion in banking. The proposed system models customers as nodes in a heterogeneous graph with relationships based on transactional similarity and communication patterns. Using a multimodal embedding strategy, the model learns customer representations via Graph Convolutional and Attention layers. Empirical results on the UCI Bank Marketing dataset demonstrate an ROC-AUC of 0.87 and accuracy of 0.886, with significant improvements over logistic regression and XGBoost baselines. Extended experiments using a heterogeneous multi-source graph (MovieLens, Last.FM, Amazon co-purchase, OGB-MAG) further confirm the framework’s superiority: accuracy 0.893 and F1-score 0.596 versus a logistic regression baseline that degenerates to F1= 0.000, AUC= 0.500. The paper details system design, dataset structure, implementation, graph construction methodology, and performance evaluation.

DOI: https://doi.org/10.5281/zenodo.20735315

Published by: vikaspatanker

Authors: Aarush Kukade, Advait Deogade, Atharva Mane, Dr. Saurabh Saoji

Abstract: In the modern digital banking era, effective marketing lead generation depends on leveraging heterogeneous, multimodal customer data. Traditional predictive models primarily rely on static, flat tabular attributes, overlooking the relational and temporal information inherent in customer transaction histories and support networks. This paper proposes MTHGT, a Multimodal Temporal Heterogeneous Graph Transformer framework that integrates multimodal data—including structured CRM attributes, sequential transactional records, and unstructured call transcripts—to predict customer lead conversion in banking. The proposed system models customers, transactions, locations, and events as nodes in a heterogeneous graph with relationships based on transactional similarity, campaign logs, and temporal history. Using a multimodal embedding strategy, the model learns customer representations via Graph Transformer layers with type-aware, distance, and temporal bias encodings. Empirical results on the Multimodal Banking Dataset (MBD; 85,620 client-month nodes, 2.26% positive rate) demonstrate that graph-based models outperform tabular baselines on ranking (HGT ROC-AUC of 0.7809 ± 0.0092 and MTHGT ROC-AUC of 0.7763 ± 0.0160 vs. Logistic Regression ROC-AUC of 0.7397 ± 0.0002). Furthermore, MTHGT improves F1-score over HGT (0.0778 ± 0.0225 vs. 0.0651 ± 0.0050) and exposes dynamic modality attributions (CRM features: 25%, dialogue text: 36%, temporal transactions: 39%), enabling explainable CRM lead scoring. The paper details system design, dataset structure, implementation, graph construction methodology, performance evaluation, and outlines a roadmap to bridge the tabular baseline gap using Focal Loss and behavioral k-NN edges.

DOI: https://doi.org/10.5281/zenodo.20735250

Published by: vikaspatanker

Authors: Dr. Saurabh Saoji, Aditya Deshmukh, Aadesh Gulumbe, Sanika Hingalkar, Akash Shelke

Abstract: Cloud-based applications increasingly rely on multiple database systems to handle diverse data models and workloads, yet managing these heterogeneous environments remains complex and resource-intensive. Traditional Database-as-a-Service platforms often introduce vendor lock-in, limited flexibility, and high costs, restricting their suitability for academic and research use. To address these challenges, this research proposes an open-source, AI-powered Cloud Database-as-a-Service platform that unifies the management of SQL, NoSQL, and in-memory databases using Kubernetes-based container orchestration. The system integrates AI-driven natural language assistance for schema generation and query formulation, along with real-time monitoring using Prometheus and Grafana. By combining automation, intelligent interaction, and cost-effective deployment, the platform aims to improve accessibility, efficiency, and scalability in cloud-native database management.

DOI: http://doi.org/10.5281/zenodo.20732646

Published by: vikaspatanker

Authors: Vishwanatha D N, Assistant Professor Jayashree K

Abstract: This research paper investigates the effect of data-driven personalization on customer engagement and brand loyalty within the digital marketing ecosystem. As organisations accumulate unprecedented volumes of consumer data through digital touchpoints—spanning e-commerce platforms, mobile applications, social media, and connected devices—the capacity to deliver highly individualised marketing experiences has grown substantially. Yet the relationship between personalization, engagement, and loyalty is complex, non-linear, and moderated by a range of consumer, contextual, and technological variables that existing literature has not yet fully integrated into a unified framework. Drawing on the Elaboration Likelihood Model (ELM), Self-Determination Theory (SDT), Relationship Marketing Theory, and the Stimulus-Organism-Response (S-O-R) framework, this paper develops a comprehensive conceptual model that traces the pathway from data-driven personalization through customer engagement to brand loyalty, incorporating personalization relevance, perceived autonomy, privacy concern, and algorithmic transparency as key moderating and mediating constructs. The paper reviews the theoretical foundations of these relationships, analyses six real-world case studies from diverse sectors including streaming, e-commerce, food delivery, and retail, and proposes a research agenda for advancing understanding of personalization dynamics in contemporary digital marketing. Key findings indicate that data-driven personalization significantly enhances customer engagement when it is perceived as relevant and non-intrusive, and that sustained engagement is the primary pathway through which personalization generates brand loyalty. However, the study also identifies critical conditions under which personalization can undermine trust and loyalty—specifically when personalisation becomes too precise, violates contextual norms, or operates without transparency. The paper concludes with strategic implications for marketers, recommendations for ethical personalization design, and directions for future empirical research.

DOI: https://doi.org/10.5281/zenodo.21773541

Published by: vikaspatanker

Authors: Ankita Patel

Abstract: Chenopodium album (Linn.), commonly known as lamb's quarters or bathua, is a fast-growing annual plant of the family Amaranthaceae with a long history of traditional medicinal use. This study investigates the antimicrobial potential of methanol and acetone leaf extracts of C. album against six pathogenic bacteria and six fungal strains using disc diffusion, well diffusion, and poisoned food techniques. Extraction was performed using the Soxhlet method with 25 g of powdered dried leaf material in 50 ml of methanol and acetone solvent mixture. Results demonstrated notable antibacterial activity against both Gram-positive and Gram-negative organisms, with acetone extract producing the largest inhibition zones against Escherichia coli (19.5 mm) by disc diffusion and 20.1 mm by well diffusion. Antifungal assays revealed that a mixture of methanolic and acetone extracts achieved up to 99% mycelial inhibition against Aspergillus niger at 7 days incubation. These findings suggest that C. album harbors broad-spectrum antimicrobial compounds with significant pharmaceutical potential.

DOI: https://doi.org/10.5281/zenodo.20732068

Published by: vikaspatanker

Authors: Rashida Noori, Atharv Patil, Samarth Gote, Om Gadre, Satish Rathod, Malik Mulani, Prof. V.P.Bhusare

Abstract: Concrete is one of the most widely used construction materials, and its compressive strength is a key parameter that determines its structural performance and durability. Traditionally, determining the compressive strength of concrete requires laboratory testing, which is time-consuming, costly, and dependent on curing conditions and sample preparation. In this study, a data-driven approach is applied to predict the compressive strength of concrete using regression analysis in Microsoft Excel. A dataset containing input variables such as cement content, water-cement ratio, fine and coarse aggregate proportions, and curing age is analysed. Various regression techniques—such as linear, multiple linear, and polynomial regression—are implemented to develop predictive models. The correlation between experimental and predicted results is evaluated using statistical indicators like R², standard error, and residual analysis. The study demonstrates that regression models can effectively predict concrete compressive strength with reasonab le accuracy, thereby reducing the need for extensive experimental trials. This approach highlights the potential of Excel as a simple yet powerful tool for engineers and researchers to perform predictive modelling and optimise concrete mix design. The compressive strength of concrete is a crucial property that determines its quality and load-bearing capacity. Conventionally, this strength is obtained through laboratory testing after curing, which can be time-consuming and resource-intensive. This project focuses on predicting the compressive strength of concrete using regression analysis in Microsoft Excel. By utilising input parameters such as cement content, water-cement ratio, fine and coarse aggregates, and curing age, a regression model is developed to estimate strength values. Multiple linear regression is applied to establish a relationship between these variables and the compressive strength. The accuracy of the model is evaluated through statistical measures like the coefficient of determination (R²) and error analysis. The results indicate that regression-based prediction provides a reliable and cost-effective alternative to traditional testing methods. This approach demonstrates the usefulness of Excel as an accessible tool for data analysis and decision-making in civil engineering applications.

DOI: http://doi.org/

Published by: vikaspatanker

Authors: Swami Siva Mahadev

Abstract: The adoption of Industry 4.0 technologies has increased the need for seamless integration between Manufacturing Execution Systems (MES) and Enterprise Resource Planning (ERP) platforms. SAP Digital Manufacturing (SAP DM), built on the SAP Business Technology Platform (BTP), provides a cloud-based solution for managing and optimizing manufacturing operations. While integration with SAP S/4HANA is supported through standardized mechanisms such as IDocs, APIs, and SAP Cloud Integration, integrating SAP DM with non-SAP ERP systems, including Oracle ERP Cloud, Microsoft Dynamics 365, and Infor CloudSuite, presents additional challenges related to data exchange, interoperability, and process synchronization. This paper proposes a unified four-layer integration framework for connecting SAP Digital Manufacturing with both SAP and non-SAP ERP systems. The framework focuses on master data synchronization, production order management, middleware architecture, security governance, and implementation strategy. By analyzing industry practices and documented integration approaches, the study demonstrates how organizations can establish a scalable and standardized manufacturing integration landscape. The paper also discusses future opportunities in event-driven architectures, artificial intelligence-based production planning, and digital twin technologies.

Published by: vikaspatanker

Authors: Ogbonna Nnamuchi

Abstract: This paper introduces a formal framework utilizing mathematical functional analysis to bridge the gap between empirical sleep science and jurisprudence. By treating fatigue trajectories as functions within infinite-dimensional Banach spaces, we formalize how biomathematical fatigue inputs intersect with duty-of-care allocations within tort and regulatory systems, shifting the legal focus from rigid shift-hour compliance to systemic accountability.

Published by: vikaspatanker

Authors: Shravani Phalke, Rajit Joshi, Raj Lohar, Bharti Dhote

Abstract: By facilitating natural, flexible, and context-aware communication across a variety of languages and cultural contexts, artificial intelligence (AI) has revolutionized human-computer interaction. Large language models have advanced, but chatbots still have difficulty identifying, interpreting, and reacting sympathetically to users' emotional states. As a result, they frequently provide generic responses that lack genuine resonance. This paper introduces Novachat, a full-stack AI chatbot designed to close this gap by combining multilingualism and sophisticated emotion intelligence into a scalable MERN-stack architecture. In order to provide human-like, contextually nuanced conversations in English, Hindi, Marathi, and other languages, Novachat's modular framework integrates sentiment analysis, emotion-adaptive response generation, and language detection. To ensure smooth real-time adaptability, each module functions as a microservice and communicates via orchestration driven by APIs. The study describes the system's overall architecture, emotional classification model, dataset organization, and quantitative performance assessment using metrics like System Usability Scale (SUS), emotion recognition accuracy, response relevancy, and user engagement latency. According to experimental results, Novachat generates sympathetic responses and detects emotions with high accuracy; a SUS score indicates strong user acceptance. The field is moving closer to AI systems that genuinely recognize and value the user's emotional experience as a result of these results, which validate Novachat's function as an efficient, inclusive, and emotionally engaging conversational platform.

Published by: vikaspatanker