EAI Endorsed Transactions on Scalable Information Systems

Grasshopper-Based Detection of Fake Social Media Profiles

Nadir Mahammed — 2025-07-24

The proliferation of fake profiles on social media platforms presents a growing challenge for digital ecosystems, where the detection of such profiles is critical to maintaining the integrity of online environments. This paper introduces a hybrid approach that integrates the Grasshopper Optimization Algorithm with various Machine Learning classifiers, including Support Vector Machine, Naive Bayes, and Random Forest. The nature-inspired metaheurisitic used is employed to optimize key hyperparameters of these classifiers, thereby enhancing their performance in detecting fake profiles. The proposed method is evaluated on a well defined balanced dataset, demonstrating significant improvements in classification performance, particularly in terms of accuracy, precision, recall, and F1-score. The results suggest that the proposed hybrid approach can effectively address the challenges associated with balanced and imbalanced datasets in fake profile detection. Furthermore, the study discusses potential directions for improving scalability and applying the approach to larger and more dynamic datasets in the future.

Predicting product sales performance using various types of customer review data

Jinthusan Baskaran — 2025-07-17

Today, in the e-commerce world, product reviews are a critical part of influencing consumer buying decisions and offer valuable insight to determine sales quality. But many current methods do not make efficient use of heterogeneous user-generated content (UGC) and those they predict with a unified model may ignore the different nature between various review types. In light of these limitations, this study introduces an integrated algorithmic framework that combines cutting-edge sentiment analyses and machine learning (ML) algorithms for sales quality prediction through automatic analysis of product reviews over the internet. The approach proposed will collect structured data from different sources during a systematic process and then consider the path of normalization, and sentiment analysis followed by feature selection to construct advanced prognosis models. The model proved highly effective, achieving an 88% accuracy rate in predicting sales quality. This strong performance indicates a significant correlation between sales performance and sentiment reviews. This new framework shows good promise that sentiment analysis in UGC can be used and deployed in e-commerce product evaluation and recommendation systems. Further research should investigate the integration of regional and temporal dynamics to improve model accuracy.

Adopting Open-Source SD-WAN: A Comprehensive Analysis of Performance, Cost, and Security Benefits Over Traditional WAN Architectures

Segun V. Arogundade — 2025-07-24

Many enterprises are using cloud computing innovation and remote services to the maximum. Working from home is becoming the norm. Favored legacy Wide Area Networks (WANs) are not up to the tasks, as they are suffering due to lack of scalability with their traditional non-virtualized form as it still requires a lot of physical components. Update and maintenance of fickle hardware costs a lot. There is a need for more flexible and scalable networking solutions. Many enterprise solutions offer proprietary form of SD-WANs (Software-Defined Wide Area Networks), but they are costly and inflexible, which means they are not practical for all applications. This paper proposes an Open-source SD-WAN with OpenDaylight platform as core that we have tested in a simulated environment along with Mininet and Oracle Virtual Box to study various scenarios. Test results show that it provides a 35% increase in throughput, decreases 40% in latency, and reduces packet loss by 50%, compared to traditional WANs. Additionally with Open-Source nature, it has a 20% lower operational coupled with the problem mitigation factors listed above, which makes it a more potential solution for the current woes of businesses.

Multimodal-Driven Emotion-Controlled Facial Animation Generation Model

Zhenyu Qiu — 2025-07-17

INTRODUCTION: In recent years, the generation of facial animation technology has emerged as a prominent area of focus within computer vision, achieving varying degrees of progress in lip-synchronization quality and emotion control.

OBJECTIVES: However, existing research often compromises lip movements during facial expression generation, thereby diminishing lip synchronisation accuracy. This study proposes a multimodal, emotion-controlled facial animation generation model to address this challenge.

METHODS: The proposed model comprises two custom deep-learning networks arranged sequentially. By inputting an expressionless target portrait image, the model generates high-quality, lip-synchronized, and emotion-controlled facial videos driven by three modalities: audio, text, and emotional portrait images.

RESULTS: In this framework, text features serve a critical supplementary function in predicting lip movements from audio input, thereby enhancing lip-synchronization quality.

CONCLUSION: Experimental findings indicate that the proposed model achieves a reduction in lip feature coordinate distance (L-LD) of 5.93% and 33.52% compared to established facial animation generation methods, such as MakeItTalk and the Emotion-Aware Motion Model (EAMM), and a decrease in facial feature coordinate distance (F-LD) of 7.00% and 8.79%. These results substantiate the efficacy of the proposed model in generating high-quality, lip-synchronized, and emotion-controlled facial animations.

VMHQA: A Vietnamese Multi-choice Dataset for Mental Health Domain Question Answering

Tu Anh Hoang Nguyen — 2025-09-25

This paper introduces VMHQA, a VietnameseMultiple-Choice Question Answering (MCQA) dataset designed to address critical mental health resources gaps, particularly in low and middle-income countries like Vietnam. The dataset comprises 10,000 meticulously curated records across 1,166 mental health subjects, including 249 topics in the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) and 8,599 contextual paragraphs. Each record adheres to the United States Medical Licensing Examination
(USMLE) format, with targeted questions, correct answers, multiple-choice options, and supporting paragraphs from reputable sources such as academic journals and local hospital websites, further inspected by prestigious mental hospitals in Vietnam. VMHQA thus provides a reliable, structured foundation for preconsultation tools, allowing for early psychological intervention for those concerned about mental health issues. This study also goes beyond data collection to evaluate the effectiveness of VMHQA using cutting-edge machine learning models, such as BERT-based architectures, large language models (LLMs) ranging from 7 to 9 billion parameters, and various generative pre-trained transformer (GPT) frameworks. In addition, we look at how Retrieval-Augmented Generation (RAG) combined with Agentic Chunking can improve the accuracy and interpretability of responses in this specialised domain. The retrieval mechanisms of RAG are examined explicitly for their ability to generate contextually accurate answers sensitive to psychological nuances. Our findings shed light on the effectiveness of these advanced models in handling complex, domainspecific question-answering tasks in mental health, highlighting their potential to make mental health care more accessible and reliable for Vietnamese-speaking communities. VMHQA thus represents a significant step toward making mental health care more accessible, offering hope for improved mental health outcomes.

User-Centric Development of Good Delivery Applications Using Design Thinking and Business Model Canvas

Johanes Fernandes Andry — 2025-10-08

INTRODUCTION: The development of internet technology drives companies to innovate and integrate technology into business processes. This study focuses on designing a goods delivery management application using a combination of Design Thinking and the Business Model Canvas (BMC). Design Thinking emphasizes user needs, while BMC ensures business sustainability. Through stages of empathizing, defining, ideating, and prototyping, this research maps user needs, business requirements, and application frameworks to address inefficiencies and modernize goods delivery processes, improving accuracy and operational effectiveness.

OBJECTIVES: This research is to design a goods delivery management application by combining Design Thinking and the BMC to align user needs with business goals, addressing inefficiencies, and improving operational processes and data accuracy.

METHODS: The research method includes five stages: literature study, data collection (empathize), problem analysis (define), ideation (solution sketch, wireframe, user flow), and prototyping. This approach combines Design Thinking and BMC to address user and business needs.

RESULTS: Defining business process inefficiencies, creating user personas and empathy maps, developing user journey maps and information architecture, designing wireframes and user flows, and proposing a Business Model Canvas (To Be).

CONCLUSION: The study highlights inefficiencies in goods delivery processes and proposes a new BMC strategy, emphasizing user-centric application development, real-time updates, resource allocation, and modular features, with future recommendations focusing on user experience enhancement and application evaluation.

Blockchain-Enabled Anti-Corruption Frameworks for Public Procurement: A Latin American Case Study

Gabriel Silva Atencio — 2025-10-02

This research examines the potential application of blockchain technology in combating corruption in public procurement in Latin America, with a focus on Costa Rica's SICOP system. The research identifies systemic weaknesses in current processes and proposes a phased framework for the adoption of blockchain. It does this by using an exploratory qualitative approach that combines ethnographic case studies of procurement officials with a comparison of how things are done in other countries. Key results show that blockchain can fill in important gaps in bid verification, audit efficiency, and process transparency. They also show that there are cultural and organizational impediments to deployment that people don't realize. The study adds to the state of the art by (1) creating a context-specific implementation model that has been tested against problems in the region; (2) measuring how well blockchain works to fight corruption through real-world benchmarks; and (3) making policy suggestions for mixed technical-institutional reforms. The results show that blockchain's worth comes not just from its cryptographic capabilities but also from its capacity to make transparency a standard way of governing. However, for it to be effective, there has to be equal investment in modernizing the law, teaching people about ethics, and getting stakeholders involved. The study goes beyond just talking about theories by giving emerging economies useful information that takes into account both technology and human components in anti-corruption initiatives.

AI-Enabled Tools: Shaping the Future of Technology

Venkata Rama Padmaja Chinimilli — 2025-10-09

INTRODUCTION: AI-enabled tools are revolutionizing various fields, ushering in a transformative era for technology and industry. These advancements impact diverse sectors, from customer service and industrial design to cybersecurity and computer vision, reshaping human-computer interactions.
OBJECTIVES: This article explores the applications and advancements of AI tools, focusing on chatbots, image and text generation, literature review research, AI-driven coding tools, and emerging applications in cybersecurity and media analysis. It also provides an overview of publicly available AI tools and conducts a comparative analysis of leading chatbots.
METHODS: A comprehensive review and analysis were conducted on tools including ChatGPT, CodeGPT, and AI systems for image and text generation. The study also examined AI applications in spear phishing defense, facial recognition across age variations (FaceNet), and deepfake detection, alongside a comparative analysis of chatbots such as ChatGPT, Google Bard, LLaMA, and MS Bing.
RESULTS: The exploration revealed that chatbots like ChatGPT have redefined customer service, while AI tools for image generation impact art, medical imaging, and industrial design. AI-driven text generation and coding tools enhance content creation and software development efficiency. Additionally, AI applications in cybersecurity, facial recognition, and deepfake detection demonstrate the technology’s growing societal relevance. Comparative analysis of chatbots highlighted their distinct capabilities across platforms.
CONCLUSION: AI-enabled tools are shaping the future of technology, driving innovation, and expanding possibilities across industries and societal domains. The findings emphasize the need for continued exploration, ethical application, and responsible deployment to maximize their potential while addressing associated challenges.

NOMA Assisted Energy-Efficient MEC for Environmental Severity Monitoring in Power IoT Networks

Guangmao Li — 2025-07-15

This paper proposes an energy-efficient mobile edge computing (MEC) scheme that utilizes non-orthogonal multiple access (NOMA) for environmental severity monitoring in Power Internet of Things (IoT) networks. The primary objective of the proposed approach is to optimize energy consumption while ensuring tasks are completed within their respective deadlines and meet reliability constraints. The scheme integrates NOMA's superposition coding with mobile edge computing to improve task offloading efficiency and reduce computational delays. To achieve this, an iterative water-filling (IWF) algorithm is applied to dynamically adjust the power allocation for each task based on varying channel conditions and latency requirements. The optimization problem is formulated to minimize energy consumption while respecting the given constraints, including outage probability and transmission rate. Simulation results demonstrate that the proposed IWF-based method significantly outperforms traditional schemes. For instance, under a stringent delay threshold of 10 ms, the IWF method reduces energy consumption by approximately 30\% compared to conventional approaches. Furthermore, even as the delay threshold increases, the IWF method consistently maintains a noticeable advantage, achieving up to 20\% lower energy consumption compared to other schemes.

Leveraging Relation Attention Mechanisms for Enhanced Knowledge Graph Completion with Embedding Translation

Jiahao Shi — 2025-10-07

In this paper, we propose a novel knowledge graph completion framework to leverage a relation-specific attention mechanism integrated with an embedding translation strategy to improve the accuracy and contextual understanding of link prediction tasks. Unlike traditional models that rely on fixed transformation spaces, the proposed method dynamically captures fine-grained relational semantics by combining hierarchical candidate categorization, relation-guided entity projection, and asymmetric score functions. Specifically, the proposed model employs K-means clustering and principal component analysis (PCA) to identify semantically consistent entity sets, and integrates attention-weighted multi-attribute embeddings to construct robust relational representations. A margin-based ranking loss with normalized embedding constraints ensures effective optimization, further supported by Xavier initialization and stochastic gradient descent. Extensive experiments on two benchmark datasets, WN18 and FB15K, demonstrate the superiority of the proposed method. Specifically, on WN18, the proposed method achieves the lowest mean rank (MR) of 144, with competitive results in mean reciprocal rank (MRR) (0.902), Hits@1 (89.0%), Hits@3 (90.4%), and Hits@10 (96.3%), closely rivaling state- of-the-art models like QuatE and ComplEx. On FB15K, the proposed method again delivers the best Mean Rank of 21, along with strong scores in MRR (0.831), Hits@1 (72.2%), Hits@3 (88.4%), and the highest Hits@10 (92.5%) among all compared methods.