HALLUCINATION IN LARGE LANGUAGE MODELS: CHARACTERIZATION, DETECTION, AND MITIGATION APPROACHES

Meenal Vardar; Mayank Sharma; Dimpal Agrawal; Ankur Vashistha; Banti Sharma; Prashansha Chaudhary; Aditya Chaturvedi; Ruchi Agrawal; Yagyavalkya Sharma; Dr. Parul Trivedi; Hemlata Bhatt; Dr. Rahul Shivaji Adnaik; Dr. Neetu Pandey; Wankasaki Lytand; Ritu Singh; Praveen Kumar; Yagyavalkya Sharma; Praveen Kumar; Dr. Neetu Pandey; Vipin Saini; Mradul Saini; Ruchi Malhotra; Yagyavalkya Sharma; Dr. Sandeep Rout; Shivani Mahura; Dr. M Ashok Kumar; Kirti Das; Pawan Singh; Dr. D Simhana Devi; R. Saisyam; Prashansha Chaudhary; Banti Sharma; Aditya Chaturvedi; Radhika Pal; Sanskar Hajela; Shivani mahura; Priyanshu Gautam; Sanskar Hajela; Shivani mahura; Mayank Sharma Mayank; Dimpal Agrawal; Kirti Sharma; Mayank Sharma Mayank; Kirti Sharma; Priyanshu Gautam; Shivani Mahura; Sanskar Hajela Sanskar; Priyanshu Gautam; Shivani Mahura; Sanskar Hajela Sanskar; Ruchi Agrawal; Yagyavalkya  Sharma; Hemlata  Bhatt; Vishal  Khandelwal

📢 Latest Update: New special issue call for papers on "Emerging Technologies in Research" - Submit by March 31, 2025

July-Dec 2025

Volume 1, Issue 1 - $2025Current Issue

Issue Details:

Volume 1 Issue 1

Published:Invalid Date

Editorial: July-Dec 2025

The inaugural issue of the International Research Journal of Scientific Reports and Reviews marks a significant step toward fostering integrity, innovation, and interdisciplinary collaboration in scientific research. This issue features pioneering studies and comprehensive reviews spanning diverse domains of science and technology, reflecting the journal’s commitment to advancing high-quality, peer-reviewed knowledge. Under the guidance of Editor-in-Chief Dr. Devendra Pratap Rao and the support of the Kalp Squad Group, this first volume sets the foundation for an inclusive global platform that bridges ideas across disciplines and promotes excellence in scientific reporting and critical analysis.

Dr. Devendra Pratap Rao
Editor-in-Chief
International Research Journal of Scientific Reports and Reviews

Articles in This Issue

Showing 11 of 11 articles

Research PaperID: IRJSRR110007

A COMPARATIVE STUDY TO MEASURE THE SUSTAINABILITY OF EXISTING RENEWABLE ENERGY SYSTEMS AND NON-CONVENTIONAL ENERGY SOURCE

Praveen Kumar, Dr. Neetu Pandey, Vipin Saini, Mradul Saini, Ruchi Malhotra, Yagyavalkya Sharma, Dr. Sandeep Rout

The global energy landscape is undergoing a crucial shift brought on by increased fossil fuel demand, environmental damage, and the need to tackle climate change. Traditional fossil fuel and nuclear energy systems have been generally dependable in the past, although there are ever-increasing worries about carbon emissions, environmental hazard, and resource availability. Renewable energy technologies including solar, wind, hydro, geothermal, and biomass represent scalable, sustainable, and low-cost alternatives as a result of technological advancements and reduced costs. Nevertheless, challenges relating to storage capabilities, intermittency of supply, and infrastructural integration remain. Furthermore, non-conventional technologies including tidal, wave, ocean thermal, micro-hydro, and second-generation bioenergy, are advancing and will provide additional diversity in energy sources in due course. This review highlights the performance, economics, social and environmental implications, and sustainability of traditional, renewable, and non-conventional energy systems. It is concluded that renewable energy represents the most immediate and practical pathway for sustainable energy transitions, while non-conventional technology can also be complementary as technologies and policies develop.

Climate change mitigationEnergy transitionnon-conventional energy sourcesRenewable energySustainable energy systems

1,770 views

486 downloads

Contributors:

Praveen Kumar

Dr. Neetu Pandey

Vipin Saini

Mradul Saini

Ruchi Malhotra

Yagyavalkya Sharma

Dr. Sandeep Rout

Research PaperID: IRJSRR110011Pages 121-138

AI-ENHANCED ADAPTIVE DROOP CONTROL FOR MULTI-MICROGRID NETWORKS UNDER HIGH RENEWABLE PENETRATION

Sanskar Hajela, Shivani mahura, Priyanshu Gautam, Sanskar Hajela, Shivani mahura

The modern multi-microgrid (MMG) systems with high renewable penetration are subject to voltage instability, inefficiency in power sharing, and long transient recovery, which are drawbacks that the conventional fixed droop controllers cannot tackle. In this context, research presents the AI-Enhanced Adaptive Droop Control (AI-ADC) framework that combines rule-based tuning with a neural-network model that can predict the optimal droop coefficients in real time. The controller is validated through a high-fidelity MATLAB/Simulink model incorporating the dynamics of PV, wind, BESS, and a converter, and subjected to load steps, irradiance fall-off, DER failure, and communication delay situations. The controller's performance with respect to voltage undershoot has been better by over 60%, the voltage steady-state deviation has been under 0.25 V, the power-sharing error has been reduced to below 3%, and the settling time has been almost four times faster than that of the conventional methods. The AI-ADC, by allowing predictive and adaptive droop adjustments, presents a scalable, efficient, and highly resilient control solution for the next-generation MMG networks dominated by renewable energy sources.

AI-Enhanced Adaptive Droop ControlMulti-Microgrid (MMG)Neural Network-Based Droop OptimisationRenewable Energy IntegrationVoltage StabilityPower Sharing

2,261 views

826 downloads

Contributors:

Sanskar Hajela

Shivani mahura

Priyanshu Gautam

Sanskar Hajela

Shivani mahura

Research PaperID: IRJSRR110013Pages 155-177

An Explainable Ensemble Machine Learning Model for Short-Term Flood Occurrence Prediction Using Hydro-Climatic Time-Series Data

Priyanshu Gautam, Shivani Mahura, Sanskar Hajela Sanskar, Priyanshu Gautam, Shivani Mahura, Sanskar Hajela Sanskar

Floods are really harmful surprise elements of nature, and as such, making short-term flood prediction very accurate is a major requirement of early-warning systems done in such a way as to prevent human and property losses, economic disruption, etc. Still, they are hard to guess due to the very intricate combination of hydrological, climatic, and other environmental factors. The present paper offers an interpretable ensemble machine learning framework for predicting the times of flood events around 1-3 days ahead via the use of hydrology, climate, and environmental indicators together. The method offers great help through data preprocessing, feature normalisation, and the application of various regression models to cost-continuous flood probabilities estimation. Random Forest and Gradient Boosting algorithms are used to find and improve prediction accuracy through capturing non-linear relationships, while a hybrid ensemble method combines the advantages of individual models. Decision-making is made simplistic by the conversion of probabilistic outputs into binary flood alerts at a predetermined fixed threshold. The framework is executed and tested in a MATLAB-Simulink setting, and the analysis confirms its readiness for real-time operations. The results from the experiments indicate that the learning through the ensemble approach has significantly improved the prediction reliability and interpretability as compared to single model techniques.

1 IntroductionIn recent yearsmachine learning ML techniques have been drawing more and more attention as the effective alternative for predicting floods due to their datadriven nature and ability to model complicated nonlinear relationships 6 Machine learning modelsunlike traditional onesat once can reveal significant patterns from large data sets without needing to rely on any specific physical assumptionsthus+4 more

2,817 views

715 downloads

Contributors:

Priyanshu Gautam

Shivani Mahura

Sanskar Hajela Sanskar

Priyanshu Gautam

Shivani Mahura

Sanskar Hajela Sanskar

Research PaperID: IRJSRR110004

CYBERSECURITY RISKS AND CONSUMER TRUST IN E-COMMERCE PLATFORMS: A BUSINESS ANALYTICS PERSPECTIVE

Banti Sharma, Prashansha Chaudhary, Aditya Chaturvedi

Cybersecurity has become one of the crucial factors of consumer trust and sustainable development of e-commerce in the fast-growing digital economy. The proposed research explores how cybersecurity risks, business analytics capabilities, cybersecurity measures, consumer trust, and purchase intent relate within the framework of Indian e-commerce platforms. A quantitative research design was used to collect data on 120 regular e-commerce users in the Delhi NCR area by using a questionnaire that was structured in a closed-ended form. As the results of the statistical data analysis according to SPSS and Excel indicated, the perceived cybersecurity risk has no significant effect on consumer trust, but the business analytics and cybersecurity measures exhibit significant correlations with trust. Moreover, there was a poor relationship between consumer trust and purchase intention, which shows that pragmatic information, comprising price and convenience, is likely to dominate online purchases in comparison to affective trust. The results point to the fact that even though sophisticated analytics and cybersecurity tools contribute to the integrity of the platform, their complexity and perceived intrusiveness can unintentionally reduce user confidence. The research has been an addition to scholarly and practical knowledge because it introduces a business analytics viewpoint into the literature of cybersecurity-trust, where it is necessary to ensure openness, user-friendly, and ethical digital measures. To inform policy and strategy towards secure and trust-based e-commerce ecosystems, implications, limitations, and future research directions are discussed.

Cybersecurity RisksBusiness AnalyticsConsumer TrustCybersecurity MeasuresPurchase IntentionE-commerce+2 more

1,341 views

539 downloads

Contributors:

Banti Sharma

Prashansha Chaudhary

Aditya Chaturvedi

Research PaperID: IRJSRR110005

DETERMINATION OF BIOACTIVE COMPOUNDS THROUGH GAS CHROMATOGRAPHY-MASS SPECTROMETRY IN MORINGA OLEIFERA LEAVES

Ruchi Agrawal, Yagyavalkya Sharma, Dr. Parul Trivedi, Hemlata Bhatt, Dr. Rahul Shivaji Adnaik

Medicinal plants have been well documented for their therapeutic and nutritional significance, and among them, Moringa oleifera Lam., also referred to as the "Miracle Tree," possesses high value due to its multifaceted pharmacological activities and profuse phytochemical makeup. Identification and characterization of bioactive molecules in the petroleum ether extract from the leaves of M. oleifera were conducted in the present work utilizing Gas Chromatography-Mass Spectrometry (GC-MS) and the potential pharmacological significance was appraised. Fresh M. oleifera leaves were procured from Chaksu, Rajasthan, authenticated, shade-dried, and pulverized, after which petroleum ether extraction was carried out through maceration. The resulting extract was subjected to GC-MS analysis and the constituents were detected by matching the retention times and mass spectra with the NIST library database. GC-MS profiling indicated a number of bioactive molecules such as stearic acid esters derivatives, carvacrol, heterocycles with boron, and bisphenol A, which possessed pharmacological activities such as antimicrobial, antioxidant, hypocholesterolemic, anticancer, and cardioprotective activities. Detection of bisphenol A stands out in view of possible environmental degradation and points up the requirement for strict quality assurance in medicinal plant research. In totality, the result verifies the fact that M. oleifera leaves are a superior storehouse of bioactive molecules with possible medicinal usage; whereas molecules such as carvacrol and stearic acid esters justify the folk usage of the plant, the detection of the toxin bisphenol A points up the importance of stringent phytochemical and safety analysis.

Gas Chromatography-Mass SpectroscopyDrumstickBioactive CompoundMoringa OleiferaMoringaceae

1,649 views

415 downloads

Contributors:

Ruchi Agrawal

Yagyavalkya Sharma

Dr. Parul Trivedi

Hemlata Bhatt

Dr. Rahul Shivaji Adnaik

Research PaperID: IRJSRR110003

HALLUCINATION IN LARGE LANGUAGE MODELS: CHARACTERIZATION, DETECTION, AND MITIGATION APPROACHES

Meenal Vardar, Mayank Sharma, Dimpal Agrawal, Ankur Vashistha

A significant barrier to preserving factual accuracy and dependability in AI-generated outputs is hallucination in large language models. Using a benchmark Kaggle dataset, this work provides a comprehensive evaluation of both advanced transformer-based architectures and traditional machine learning classifiers for hallucination identification. They compared refined transformer models, such as DistilBERT, RoBERTa, and DeBERTa, with baseline models, including Random Forest, SVM, and Logistic Regression. The results show that transformer-based models were more robust and better at understanding context; however, more conventional models, such as Random Forest, achieved a high overall accuracy of 94.10%. DistilBERT struck a wonderful balance between precision and readability. The confusion matrix analysis demonstrated that the models helped reduce false alarms for non-hallucination outputs. The ROC-AUC ratings confirmed the transformers’ precision and capability for identifying a slight rate of semantic discrepancies. Other studies provided supporting evidence that deeper context modeling will provide real benefits to the reliability of detection rates, demonstrated by the reduced hallucinations and assessments of the frequency of errors made. In conclusion, this research shows that combining traditional and modern approaches is beneficial and that tuning with transformer models holds promise for reducing hallucinations. This research provides an example of early steps of increasing trustworthiness and human-like models as AI models.

Hallucination DetectionLarge Language ModelsTransformer-Based ModelsMachine LearningTrustworthy AI

1,334 views

506 downloads

Contributors:

Meenal Vardar

Mayank Sharma

Dimpal Agrawal

Ankur Vashistha

Research PaperID: IRJSRR110010Pages 108-121

Impact of Social Commerce Analytics on Consumer Buying Patterns in Indian E-Retail

Prashansha Chaudhary, Banti Sharma, Aditya Chaturvedi, Radhika Pal

The research paper examines how Social Commerce Analytics (SCA) operates as a driving force behind changes in consumers' perceptions in the Indian e-retailing market. It outlines the interrelatedness of SCA, drinking, consumer trust, personalization, and engagement metrics in consumer buying behavior. The research was quantitative, and the number of participants was 100, on whom correlation and regression analyses were performed using SPSS to test the three hypotheses. The results revealed a weaker, yet statistically significant, negative correlation between SCA effectiveness and consumer buying behavior, indicating a gap between SCA capabilities and consumer expectations. Likewise, consumer trust was found to have a weak negative correlation with purchase intention, indicating that factors like convenience and pricing might more heavily influence the purchase decision than trust. Conversely, personalization and engagement metrics were highlighted as very positively influential to both purchase frequency and brand loyalty, which was statistically significant (R = .708, p < .001) and accounted for 50.1% of the variance. This suggests that data-driven personalization can attract consumer engagement and loyalty for a long time. This study further broadens the discussion on social commerce from the perspective of emerging markets, and on top of that, it provides valuable managerial implications for improving consumer experience by means of data analytics that are appropriate, ethical, and effective.

Social Commerce AnalyticsConsumer Buying PatternsIndian E-RetailConsumer TrustPurchase IntentionPersonalization+2 more

2,307 views

607 downloads

Contributors:

Prashansha Chaudhary

Banti Sharma

Aditya Chaturvedi

Radhika Pal

Research PaperID: IRJSRR110017

MICROBIAL ECOLOGY OF DENITRIFICATION PROCESS AND ITS APPLICATION IN WASTEWATER TREATMENT: CHALLENGES AND OPPORTUNITIES

Dr. Neetu Pandey, Wankasaki Lytand, Ritu Singh, Praveen Kumar, Yagyavalkya Sharma

Denitrification is a critical microbial process for nitrogen removal in wastewater treatment, offering a cost-effective and sustainable alternative to conventional chemical and physical methods. This review synthesizes current knowledge on the microbial ecology of denitrification, focusing on the diversity, physiology, and community dynamics of denitrifiers in biofilms and activated sludge systems. Key bacterial genera, including Pseudomonas, Paracoccus, Hyphomicrobium, Comamonas, and Azoarcus, play dominant roles, with carbon sources such as methanol, ethanol, acetate, and waste-derived substrates strongly shaping community structure and function. Advances in molecular approaches—such as PCR-based techniques, stable isotope probing, fluorescence in situ hybridization, metagenomics, and transcriptomics—have provided new insights into microbial diversity, gene expression, and metabolic pathways, linking ecological patterns with treatment performance. Applications of denitrification span conventional activated sludge processes, biofilm reactors, and emerging autotrophic methods such as anammox, which enhance nitrogen removal efficiency. Despite these advances, operational challenges remain, including incomplete denitrification, seasonal failures, greenhouse gas emissions, and limited predictability of microbial responses to environmental shifts. Integrating molecular data into process models and optimizing carbon source utilization represent key strategies for future improvement. This review highlights the opportunities and challenges in bridging microbial ecology with engineering practices, ultimately advancing wastewater treatment technologies toward greater sustainability and resilience.

DenitrificationMicrobial EcologyWastewater TreatmentCarbon SourcesMolecular Techniques

1,568 views

490 downloads

Contributors:

Dr. Neetu Pandey

Wankasaki Lytand

Ritu Singh

Praveen Kumar

Yagyavalkya Sharma

Research PaperID: IRJSRR110009

Microscopic Evidence of Flavonoid Accumulation in Specific Stem Tissues of Maytenus senegalensis

Ruchi Agrawal, Yagyavalkya Sharma, Hemlata Bhatt, Vishal Khandelwal

The present study was carried out to check the presence of flavonoid in the stem of Maytenus senegalensis by using microscopic technique. It revealed the presence of flavonoid in stem by using histochemical tests. This study is helpful in the process of quality control and authentication of the plant in the taxonomy and for scientific classification.

Quality controlFlavonoidsHistochemical analysisAnatomy of stem

1,881 views

579 downloads

Contributors:

Ruchi Agrawal

Yagyavalkya Sharma

Hemlata Bhatt

Vishal Khandelwal

Research PaperID: IRJSRR110008

PROCESS–DEFECT–PERFORMANCE RELATIONSHIPS IN ADDITIVE MANUFACTURING OF AEROSPACE ALLOYS: A CRITICAL REVIEW OF ADVANCES AND CHALLENGES

Shivani Mahura, Dr. M Ashok Kumar, Kirti Das, Pawan Singh, Dr. D Simhana Devi, R. Saisyam

Additive manufacturing (AM) has advanced considerably as a disruptive technology that addresses the increasing demand for multi-functional, multi-material, and geometrically complex components and has begun to reshape areas of both product development and production methods. However, significant challenges related to incompatible material properties, defective parts made from AM processes, and inconsistent use of product build quality still exist. The ability to achieve highly precise AM processes is heavily reliant on monitoring, controlling, and utilizing multiple process variables effectively. In recent years, more advanced methods, including machine learning (ML), big data analytics, and design for additive manufacturing (DfAM), have become more available to address these limitations. Although these methods have been studied significantly, their use in aerospace-specific settings has been limited. In this review, study first provides an in-depth discussion on recent trends in DfAM. The review also considers simulation and modeling tools as enablers to obtain improved geometric fidelity in AM and to increase predictability of AM processes. Other trends include improved automation through the IoT, as well as knowledge-based process planning, also for multi-part cases of production. Lastly, review summarize many of the on-going issues and some future directions for algorithm-driven AM in aerospace and relate to industry 4.0 trends that emphasize intelligent automation, adaptive process control and lifecycle management, which will improve efficiency, reliability, and scale-up of AM technologies in aerospace and other advanced engineering applications.

Additive ManufacturingAerospaceBig Data AnalyticsDfAMIoTMachine Learning

1,738 views

684 downloads

Contributors:

Shivani Mahura

Dr. M Ashok Kumar

Kirti Das

Pawan Singh

Dr. D Simhana Devi

R. Saisyam

Research PaperID: IRJSRR110012Pages 139-154

Robust and Structure-Aware Visual Representation Learning for Reliable Deep Neural Networks

Mayank Sharma Mayank, Dimpal Agrawal, Kirti Sharma, Mayank Sharma Mayank, Kirti Sharma

The focus of this study's strong and structure-aware visual representation learning framework is medical picture analysis, which aims to make deep neural networks more reliable, resilient, and easy to understand. To transcend accuracy-focused evaluation, edge-guided structural oversight, corruption-sensitive robustness assessment, and calibration-oriented reliability analysis are introduced. The structure-aware MobileNetV3 does well on the Chest X-Ray dataset, with an accuracy of 0.8574, a high average confidence of 0.9284, and a controlled Expected Calibration Error (ECE) of 0.0710. The structure-aware ResNet-18 achieved an accuracy of 0.9071 and a low ECE of 0.0160. DenseNet121 had an accuracy of 0.8894 and an ECE of 0.0319. A robustness study reveals that performance trends remain consistent with ROC-AUC values exceeding 0.92, even after multiple changes, including the presence of Gaussian noise and occlusion. Grad-CAM explainability analysis demonstrates an anatomically directed emphasis on pulmonary regions, reinforcing structural priors. To evaluate the system's ability to work outside of medical imaging, it is tested on CIFAR-10. The robust model gets 71.92% clean accuracy, and this number goes up a lot when noise and blur corruptions are included. The results indicate that structure-aware and reliability-driven learning enhances the behavior of trustworthy models, making the proposed framework appropriate for real-world, safety-critical visual recognition systems.

Structure-Aware Representation LearningRobust Deep Neural NetworksReliability and Calibration AnalysisExplainable Medical Image ClassificationCross-Domain Robustness Evaluation

2,344 views

728 downloads

Contributors:

Mayank Sharma Mayank

Dimpal Agrawal

Kirti Sharma

Mayank Sharma Mayank

Kirti Sharma