Journal of Computer Science Engineering and Software Testing

Advancing Language Model Intelligence through Retrieval-Augmented Strategies

2025-09-11T10:59:20+00:00

Retrieval-Augmented Generation (RAG) is a groundbreaking concept that has transformed the enhancement of Large Language Models (LLMs) by addressing significant challenges such as providing inaccurate information, experiencing hallucinations, and lacking sufficient knowledge. By combining the ability to retrieve external knowledge stores and their generative modeling, RAG offers a way in which LLMs can dynamically consult and ingest pertinent knowledge based on up-to-date document sets when generating responses. It takes the best of both worlds since it leverages information retrieval techniques, yet its natural language generation with the help of deep learning, producing output, which is not only of high quality and based on real data but is also contextually specific and rich. In RAG systems, two stage pipeline is more common, with a retriever finding relevant documents upon user query, and a generator making an intelligible response using user query and the results retrieved. This architecture equips LLMs with the means to work well in areas where domain-specific information is necessary, in real-time scenarios, or with long contexts, where the pre-trained models cannot perform. In addition, RAG simplifies interpretation by ensuring that sources of content produced by AI can be traced, bringing more trust to AI-based systems. The range of applications is quite broad, covering such domains as healthcare, finance, law, and customer service, where precision and reliability are crucial. With the ongoing development of LLMs, RAG serves as a significant step towards the unification of the distinction between predetermined model parameters and the infinite number of human knowledges. Future applications include maximizing the quality of retrieval, latency reduction, augmented context fusion, and consideration of multimodal elements, with the ability to perform those functions as a key step in developing more intelligent, capable, and sensitive AI systems.

Design and Development of an Online Pizza Delivery System

2025-11-04T10:35:07+00:00

This paper presents the design and implementation of a web-based online pizza delivery application using Java Spring Boot, JSP, and MySQL. The platform enhances the pizza ordering process by integrating real-time order tracking, user authentication, a dynamic menu system, secure online payment processing, and an administrative control panel. With a focus on scalability and maintainability, the system adopts the Model–View–Controller (MVC) architecture, ensuring smooth interaction between the front end and back end. Experimental results show a high success rate for user registrations and order placements, validating the system’s effectiveness. This application is particularly beneficial for small-scale pizza outlets looking to digitize their operations without relying on third-party platforms.

Study on Resource Access Management Method Based on Entropy Weighting Analysis in SSO System

2025-12-09T06:39:19+00:00

The primary challenge in the current information protection system is to correctly authenticate users who access information systems and to establish the corresponding access management system. Recently, the SSO (Single Sign On) system has been developed to provide various services with a single subscriber authentication across various service systems, and its application has been intensified, with research efforts made to improve access control by comprehensively analysing the user system behaviour. SSO system can be a great risk for all service systems if it applies access control systems that are not secure due to their characteristics. The flexibility and security of access control for service systems are currently being achieved through the degree of user confidence worldwide. This study uses entropy weighting to quantitatively and accurately evaluate user behaviour, and based on it, it proposes an efficient resource access management method for the SSO system and evaluates its performance, solving the authentication and access control problem that exists in the SSO system.

Improving Credit Card Fraud Detection through Multi-model Deep Learning: A Comparative Study of ANN and VGG Networks

2025-12-11T08:38:37+00:00

The discovery of credit card fraud is still very hard because fraudsters are getting smarter, and transaction data is very unequal between classes. This research shows a clever approach that integrates three advanced models—an Artificial Neural Network (ANN) enhanced with batch normalization and dropout, and the VGG16 and VGG19 architectures—to make detection more accurate and reliable. After a lot of data preprocessing, the system starts up. It uses standard scaling to normalize the data, the Synthetic Minority Over-sampling Technique (SMOTE) to ensure balanced distribution, and Principal Component Analysis (PCA) to redundant dimensions. The cleaned dataset is fed into three separate models—a deep ANN with sequential dense layers (256 → 128 → 64 → 32) using ReLU activation, batch normalization for training stability, and dropout (rate = 0.3) to prevent overfitting, as well as VGG16 and VGG19, which are specially designed to operate on reshaped transaction data and extract complex and nuanced fraud-related features through their deep convolutional layers. An analysis of performance shows that the ANN model does a better job than VGG16 and VGG19, which have accuracy rates of 94% and 95%, respectively. It gets results that are 99% spot on. This study compares convolutional models to ANNs and shows that both are useful for finding scams. The suggested method offers an adaptable and effective way to find credit card fraud in real time.

AI-generated Deepfakes in the Age of Misinformation: A Review of Methods, Impacts, and Defenses

2025-12-23T12:27:37+00:00

The presence of artificial intelligence technologies, particularly those that enable the creation of hyper-realistic synthetic media, also known as deepfakes, has introduced serious challenges to information integrity. AI systems utilize deep learning techniques like generative adversarial networks (GANs) to generate sound, video, or images that would be capable of convincingly imitating real people and events. Although these technologies can potentially be applied to benefit entertainment, education, and accessibility, they are extremely dangerous when used inappropriately. Deepfakes can be used to produce misinformation, accidentally disseminate false information, and disinformation, intentionally misleading information to sway opinion or ruin reputations. The potential deepfakes hold in undermining media credibility, fueling political and social tensions, and enabling fraud or manipulation of identity is a reason why effective detection systems, ethics, and legislation need to be in place. This is an overview of the dual-use character of deepfake technology, technical foundations, its social impact, and existing mitigation and regulation.