Journal of Science Engineering Technology and Management Science

The exponential growth of social media has led to the continuous generation of vast amounts of usergenerated textual content, necessitating efficient automated techniques for meaningful analysis. The release of The Social Dilemma triggered extensive global engagement, producing a large volume of tweets that capture varied public perspectives on issues such as digital ethics, platform governance, and regulatory concerns. Traditional manual analysis methods are inadequate for handling such large-scale and dynamic datasets due to limitations in time, efficiency, and scalability. To overcome these challenges, this study presents an integrated Natural Language Processing (NLP) framework designed for concurrent sentiment detection and topic categorization. The workflow begins with structured preprocessing, including tokenization, stop word elimination, and lemmatization, followed by Exploratory Data Analysis (EDA) to identify underlying textual patterns and distribution trends. Context-aware semantic features are extracted using RoBERTa-based embeddings, enabling a richer representation of textual meaning. To address data imbalance issues, the Synthetic Minority Oversampling Technique (SMOTE) is incorporated, ensuring fair representation across different classes. The processed features are then used to train multiple baseline classifiers, such as Decision Tree (DT), K-Nearest Neighbor (KNN), and Naïve Bayes (NB), facilitating comparative performance analysis. Furthermore, a Deep Neural Network (DNN) is utilized as a feature extractor, and its learned representations are enhanced through a Tree Alternating Optimization (TAO) Tree Classifier. The proposed Social Transform Deep Tree (STDT) framework effectively categorizes sentiments into Negative, Neutral, and Positive, while also identifying key thematic topics.

Keywords : Sentiment Analysis, Topic Classification, Natural Language Processing (NLP), Social Media Analytics, Text Classification, Social Transform Deep Tree (STDT).

Author : K. Anusha Reddy, K. Vamshee Krishna, Chinthakindi Koushika, Racherla Harshith, Shivuni Ganindra

Title : Robust RoBERTa-Based NLP Framework for Dual-Target Sentiment and Topic Classification in Social Media Discourse

Volume/Issue : 2026;03(04(1))

Page No : 1-11

In contemporary organizations, employee reviews shared on platforms such as Amazon represent a valuable source of insight into workplace dynamics; however, deriving meaningful information from this largely unstructured textual data remains a significant challenge. Conventional approaches, including manual inspection, heuristic-based analysis, and simple keyword-driven sentiment techniques, are not only time-intensive and inconsistent but also inadequate for capturing the nuanced context and multiple dimensions embedded within employee feedback. As a result, organizations face difficulty in accurately assessing critical workforce factors such as work-life balance, career progression, and job security when dealing with large volumes of reviews. These traditional methods lack scalability, struggle with complex sentiment interpretation, and often yield suboptimal accuracy, underscoring the necessity for a more advanced and intelligent solution. To overcome these limitations, the proposed system employs a transformer-based Natural Language Processing (NLP) framework integrated within a real-time Flask web application, utilizing Sentence-BERT (SBERT) embeddings for deep semantic feature extraction alongside a Stochastic Gradient Descent (SGD) based multi-label classification model, with a Hashing Vectorizer combined with Bernoulli Naive Bayes (BNB) serving as a comparative baseline. The framework executes a comprehensive pipeline encompassing data preprocessing, feature engineering, model training, prediction, and evaluation, enabling the simultaneous identification of multiple workplace experience indicators from a single review. By leveraging contextual understanding and scalable automation, this approach enhances prediction accuracy, minimizes manual intervention, and delivers actionable insights, thereby supporting datadriven human resource analytics and informed organizational decision-making

Keywords : Employee Reviews, Workplace Dynamics, Unstructured Text Analysis, Sentiment Analysis, Work-Life Balance, Career Progression, Job Security, Human Resource Analytics, Textual Data Processing

Author : K. Sunil Kumar,M. Ramana Kumar,Ammana Saketh reddy, Guguloth Santhosh, Dubbaka Saikiran

Title : A Transformer-Based Multi-Label NLP Architecture for Extracting Workforce Experience Indicators from Amazon Employees Data

Volume/Issue : 2026;03(04(1))

Page No : 12-23

Autonomous vehicular communication systems play a vital role in intelligent transportation by supporting continuous data exchange among vehicles, roadside infrastructure, and centralized networks. These systems depend on key parameters such as Random Access Memory (RAM), storage capacity, transmission rate, and trust factor to ensure efficient and reliable communication. In dynamic environments, accurate evaluation of communication unit capability is essential for maintaining performance, safety, and optimal resource utilization. Conventional assessment methods rely on manual configuration checks and thresholdbased monitoring, where parameters are evaluated individually. Such approaches fail to capture the complex relationships among multiple factors, making them unsuitable for real-time vehicular scenarios and leading to inaccurate capability estimation. To improve automation, machine learning models including Decision Tree Regressor (DTR), Orthogonal Matching Pursuit Regressor (OMPR), and K-Nearest Neighbors Regressor (KNNR) are used to predict capability scores. However, these models have limitations in handling nonlinear interactions, noise sensitivity, and generalization. To address these issues, a hybrid Deep Fuzzy Regression (DFR) model is proposed, integrating Deep Fuzzy Encoding (DFE) with Random Forest Regressor (RFR) and Linear Regression (LR) through ensemble learning. The system follows a pipeline of preprocessing, feature handling, training, and evaluation using MAE, MSE, RMSE, and R². Results show that DFR provides accurate and reliable capability assessment for real-time vehicular communication systems.

Keywords : Autonomous vehicular communication, intelligent transportation systems, vehicular networks, communication capability assessment, resource optimization, transmission rate, memory management.

Author : K. Anusha Reddy, Pabbathi Upender, Gunjolu Sravani, Vanam Sathya Sri Asritha

Title : Cognitive Fuzzy Learning with Continuous Regression for Online Capability Inference in Vehicular Systems

Volume/Issue : 2026;03(04(1))

Page No : 24-31

Contemporary defense systems are increasingly dependent on intelligent technologies to enhance surveillance, reconnaissance, and strategic decision-making capabilities. The rapid growth of visual data generated from satellites and unmanned aerial platforms has made manual image analysis and traditional rule-based approaches inefficient, labor-intensive, and unsuitable for large-scale processing. To overcome these limitations, the proposed system employs advanced Machine Learning and Deep Learning techniques to automate the classification of military imagery. The framework integrates multiple algorithms, including Perceptron, Decision Tree Classifier (DTC), Deep Neural Networks (DNN), and a Hybrid Convolutional Recurrent Model (CRM). The CRM combines Convolutional Neural Networks (CNN) for robust spatial feature extraction with Long Short-Term Memory (LSTM) networks to model complex sequential relationships within the extracted features. These models are trained on well-structured military image datasets to accurately categorize assets into predefined strategic classes. Experimental results demonstrate that DNN and CRM models deliver improved accuracy and reliability, depending on dataset characteristics and training conditions. Additionally, the system is equipped with a user-friendly graphical interface that allows seamless dataset uploading, model training, prediction generation, and visualization of performance metrics. By consolidating multiple models into a single platform and identifying the most effective one for deployment, the proposed solution significantly improves the efficiency, accuracy, and robustness of automated military image analysis, thereby enhancing intelligence operations and supporting informed decision-making.

Keywords : Image Analysis, Surveillance Systems, Defense Intelligence, Pattern Recognition, Automated Classification, Deep Learning.

Author : M. Ramana Kumar, Polasa Vaishnavi, Dyagala Varshith, Mohammed Sohail

Title : AI-Driven Tactical Military Asset Recognition for Enhanced Situational Awareness and Command Decision Support

Volume/Issue : 2026;03(04(1))

Page No : 32-41

The rapid evolution of digital healthcare solutions has created a critical need for systems that can manage Electronic Health Records (EHR) securely while supporting intelligent decision-making. Accurate prediction of heart disease and effective handling of patient data require both advanced analytical models and strong privacy protection, which traditional centralized systems fail to provide due to risks such as data breaches, unauthorized access, and lack of transparency. Moreover, these systems do not effectively combine predictive analytics with secure data sharing, limiting their use in real-time healthcare scenarios. To overcome these challenges, this work presents a secure healthcare framework that integrates Machine Learning (ML), blockchain technology, and Ciphertext-Policy Attribute-Based Encryption (CP-ABE). The system processes a heart disease dataset using preprocessing techniques such as missing value handling, normalization with Standard Scaler, and traintest splitting, followed by prediction using Random Forest (RF) and Support Vector Machine (SVM), with performance evaluated through accuracy, precision, recall, and F1-score. A web-based interface is developed using Django to facilitate interaction between doctors and patients. For enhanced security, patient data is encrypted using Elliptic Curve Integrated Encryption Scheme (ECIES)-based CP-ABE before being stored on the blockchain through Web3, while smart contracts manage user registration, EHR storage, and controlled access. This integrated approach ensures secure data sharing, reliable prediction of heart conditions, and transparent record management, ultimately improving data security, prediction accuracy, and enabling decentralized, tamper-proof healthcare services

Keywords : Electronic Health Records (EHR), Heart Disease Prediction, Machine Learning (ML), Blockchain Technology, Elliptic Curve Integrated Encryption Scheme (ECIES)

Author : Pulime Satyanarayana, Chandur Vedhasri, Kolugoori Gunasagar, Bhoopathi Mani Kumar

Title : Lightweight Secure and Transparent Ensemble Learning for IntegrityPreserved Health Records

Volume/Issue : 2026;03(04(1))

Page No : 42-51

Urban environments generate complex acoustic landscapes with overlapping sound events such as sirens, horns, and traffic noise, making automated monitoring for smart cities, public safety, and noise pollution control highly challenging. Traditional sound classification systems rely on hand-crafted features like Mel-Frequency Cepstral Coefficients (MFCCs) and log-Mel spectrograms, combined with models such as Support Vector Machines (SVMs), Random Forests (RF), or shallow Convolutional Neural Networks (CNNs), achieving only modest accuracy. Emerging from early 2010s research and benchmarked on datasets like UrbanSound8K, these approaches suffer key limitations: inability to capture semantic audio understanding in overlapping conditions, neglect of multi-label correlations, lack of interpretability in deep models, and dependence on command-line interfaces that limit usability for non-experts. This research proposes a Whisper-powered multi-task urban sound classification system with an integrated Graphical User Interface (GUI) built using Tkinter and secured with rolebased authentication (Lightning Memory-Mapped Database (LMDB) and SHA-256 hashing). It leverages OpenAI Whisper-base as a feature extractor, generating robust representations through mean pooling of encoder hidden states from class-organized audio data. These features are used to train four interpretable models: Boosted Rules Classifier (BRC), Hierarchical Structural (HS) Tree Classifier, Sparse Linear Integer Model (SLIM), and Marginal Shrinkage Linear Trees (MSLT), enabling dualtask classification of primary categories and subcategories. The system enhances transparency, usability, and performance by combining transformer-based representations with interpretable machine learning. It democratizes access to explainable artificial intelligence (AI) for urban monitoring, enabling secure, visual, and multi-task sound analysis for real-world smart city applications.

Keywords : Urban sound classification, acoustic scene analysis, environmental noise monitoring, overlapping sound events, smart cities, public safety, noise pollution control, semantic audio understanding.

Author : P. Satyanarayana, Thupakula Rasagna, Sahithya Ayla, Md Umer

Title : A Role-Authenticated Interpretable AI System for Multi-Tiered Urban Noise Monitoring in Smart Cities

Volume/Issue : 2026;03(04(1))

Page No : 52-62

Industry and brand classification was earlier done using manual taxonomies and rule-based systems. These methods were slow and needed a lot of human effort. They often made mistakes with similar brand names or overlapping industries. Traditional machine learning models improved speed but still needed manual feature design. They could not fully understand the meaning of words in context. With new deep learning methods, transformers brought a big change. Robustly optimised bi-directional encoder representation with transformers pretrained approach (RoBERTa) is one such model that reads text with better understanding. In this research, Robustly optimised bi-directional encoder representation with transformers pretrained approach (RoBERTa) taxanomy, uses this model for automated industry and brand classification. It reduces human work and gives more accurate results. This shows how automation with NLP is better than old manual approaches. Robustly optimised bidirectional encoder representation with transformers pretrained approach (RoBERTa) Taxonomy is an automated product classification system designed to identify the industry, brand, and category of any product text using transformer-based Natural Language Processing (NLP). This research leverages Robustly optimised Bi-directional Encoder Representation with Transformers pretrained approach (RoBERTa), a state-of-the-art transformer model, fine-tuned on labeled product data collected from titles, descriptions, and reviews. The system processes raw text through cleaning, tokenization, and multi-task classification layers that independently predict industry, brand, and hierarchical product categories. By mapping predictions into a structured taxonomy, the model provides accurate and consistent product tagging

Keywords : Industry Classification, Brand Classification, Product Categorization, Taxonomy Mapping, Natural Language Processing (NLP), Transformer Models

Author : K. Chiranjeevi, B. Poojitha, K. Balakrishna, N. Siva Nagamani

Title : RoBERTa-Based Deep Learning Approach for Industry and Brand Classification Systems

Volume/Issue : 2026;03(04(1))

Page No : 53-67

The rapid growth of the telecommunications sector has resulted in massive volumes of customer-agent interaction data, making manual sentiment analysis infeasible. This research presents a robust framework for telecom transcript sentiment detection by combining advanced Natural Language Processing (NLP) techniques, transformer-based embeddings, and ensemble Machine Learning (ML). The system begins with data preprocessing, including text cleaning, tokenization, stopword removal, and lemmatization, followed by exploratory analysis using word clouds, document length distributions, POS tagging, and bigram frequency plots to uncover textual patterns. Google Pathways Language Model (PaLM) like embeddings are then extracted using transformer models to capture rich contextual semantics, and class imbalance is addressed using Random Under Sampler for uniform representation across sentiment classes. Multiple ML models, including Logistic Regression Classifier (LRC), Decision Tree Classifier (DTC), Extra Trees Classifier (ETC), Boosted Rules Classifier (BRC), and a custom Fast Interpretable Greedy-Tree Sums (FIGS) ensemble classifier, are trained and evaluated, with the FIGS model aggregating predictions from base learners to enhance accuracy, robustness, and generalization. The framework supports real-time prediction, model persistence, and visualization of performance metrics, providing interpretable insights for telecom operations. Evaluation results, including accuracy, precision, recall, and F1-score, demonstrate the effectiveness of the proposed approach. The system offers a scalable, efficient, and interpretable solution for automated sentiment detection in telecom transcripts, enabling service providers to improve customer experience, monitor agent performance, and make data-driven operational decisions.

Keywords : Telecommunications, Sentiment Analysis, Natural Language Processing, FIGS Classifier, Real-Time Prediction

Author : R. Deepthi, N. Divya Sruthi, Shaik Sameera, Thuremerla Lahari, Siginal Neelima, Upputuru Sushma

Title : FIGS - Driven Contextual Sentiment Analysis for Large Scale Telecom Conversations

Volume/Issue : 2026;03(04(1))

Page No : 68-77

In modern railway networks, real-time monitoring and secure control communications are essential for ensuring operational safety, efficiency, and reliability. Critical operations such as signal control, train scheduling, and automated track switching generate large volumes of data that must be analyzed instantly to prevent failures or detect malicious intrusions. Traditional systems rely on manual inspections or rule-based approaches, which are slow, error-prone, and inadequate for handling the dynamic and high-volume nature of modern rail communication data. These limitations highlight the need for a robust, automated anomaly detection framework capable of accurate real-time classification. Existing methods, including Decision Tree with Cost Complexity Pruning (DTCCP) and Deep Neural Decision Tree (DNDT), provide interpretable models with moderate predictive performance. However, DTCCP often suffers from overfitting when dealing with complex sequential data, while DNDT may struggle to capture subtle contextual relationships, resulting in missed anomalies or false alarms. To address these issues, this research proposes a RuleFit (RF) classifier that combines linear rules with decision tree logic and semantic embeddings derived from Sentence-BERT (SBERT). This hybrid approach enables the system to learn both hierarchical decision boundaries and contextual patterns effectively. Performance evaluation using metrics such as accuracy, precision, recall, F1-score, confusion matrix, and ROC curves demonstrates that the proposed method significantly improves anomaly detection while reducing false positives and supporting reliable, real-time rail communication monitoring

Keywords : Railway Communication Security, Real-Time Monitoring, RuleFit Classifier (RF), Intelligent Transport Systems (ITS), Cybersecurity in Rail Networks

Author : Sk. Asiff, R. Deepthi, Shaik Ishaq, Shaik Abdul Bhasha, Konisetty Manoj Kumar, Yakasiri Penchala Prasad

Title : Context-Aware Rule Ensemble Learning for Adaptive Threat Detection in Intelligent Railway Signaling Data Streams

Volume/Issue : 2026;03(04(1))

Page No : 78-87

Crop disease detection is a critical challenge in modern agriculture that directly impacts food security and crop yield worldwide. This Project presents an embedded system for real-time crop disease detection using a Raspberry Pi microcomputer integrated with a USB camera and a deep Convolutional Neural Network (CNN) model trained on a large dataset of diseased and healthy plant images. The proposed system captures images of plant leaves in the field, preprocesses them, and feeds them through the CNN model to classify the type of disease with high accuracy. Results obtained from the system are displayed on a connected LCD screen and simultaneously transmitted to a remote user via a Telegram Bot for immediate notification. making it a cost-effective and practical solution for real-time agricultural disease monitoring and early intervention.

Keywords : Crop Disease Detection, Convolutional Neural Networks, Raspberry Pi, Deep Learning, Image Classification, Telegram Bot, LCD Display, Embedded Systems, Plant Pathology, Precision Agriculture

Author : G. Udaykiran Bhargava, B Tejaswi, L Hemanth, B Dharma Teja, B Sai Chandhu, Sk Aman

Title : CROP DISEASE DETECTION USING RASPBERRY PI CAMERA AND DEEP CNN’S

Volume/Issue : 2026;03(04(1))

Page No : 88-95

Rescuing individuals trapped in deep, narrow borewells is an extremely complex task due to limited space, absence of light, and the impracticality of direct human involvement. This research presents a compact and wirelessly controlled Borewell Rescue Robot (BRR) designed to improve safety and efficiency during such emergencies. The system is built around an ESP32-CAM module, which combines processing capability with onboard imaging to deliver real-time video feedback from inside the borewell. The robot is capable of transmitting live video using a lightweight communication protocol, enabling operators to monitor conditions and guide actions with minimal delay. A pair of hightorque servo motors forms the core of the mechanical system, allowing precise movement and controlled gripping of the victim within confined vertical shafts. To address visibility challenges, an adjustable Light Emitting Diode (LED) lighting unit is integrated, providing sufficient illumination even in completely dark environments. The control system is developed using a non-blocking, eventdriven architecture that allows simultaneous handling of motion commands and video streaming. This ensures smooth operation without lag or interruption. The robot functions through a self-contained wireless network, making it suitable for deployment in remote or infrastructure-limited areas. To enhance reliability, a safety mechanism is included that automatically stabilizes the system and halts movement if communication is disrupted. This solution offers a practical, scalable, and affordable approach to improving rescue operations in hazardous underground situations.

Keywords : Borewell Rescue Robot (BRR), ESP32-CAM MCU, Real-Time Video Streaming, Emergency Robotics, WebSocket Communication

Author : Md. Sharmila, P. Tejaswini, B. Nandini, P. Aravindh, M. Shiva Parvathi, B. Vijay Kumar

Title : A Low-Latency Visual Feedback Driven Robotic Mechanism for Rescue Operations in Restricted Vertical Structures

Volume/Issue : 2026;03(04(1))

Page No : 96-104