Modern pull request (PR) review is increasingly constrained by repository scale, architectural complexity, and governance expectations. Prior studies show that review effectiveness depends on change understanding and reviewer context [1], while architecture-consistency research shows that implemented structures frequently drift from intended designs during software evolution [2], [3]. This paper presents AI-Based-Quality-Check-On-Project-Code-And-Architecture, a graph-augmented review framework that couples GitHub event ingestion, abstract syntax tree (AST) parsing, dependency-graph modeling, and configurable large language model (LLM) reasoning with governed project control. The framework (i) enriches pull request assessment with structural and repository context, (ii) detects circular dependencies, cross-layer violations, and coupling anomalies before they become deeply embedded in routine development, and (iii) records every review execution in an auditable, role-controlled workflow. A containerized prototype integrating a Next.js frontend, a FastAPI backend, PostgreSQL, Neo4j, and Redis provides a reproducible baseline for studying how AI-assisted review, graph-based architecture analysis, and software quality governance can be combined within a single engineering workflow.

Software quality cost management is a critical challenge in modern development teams, where quality-related data is typically fragmented across multiple tools such as project management systems, defect trackers, and time-logging sheets. This paper presents COQ MONITOOL, a full-stack web application designed to centralize and automate the tracking and analysis of software quality costs using the Prevention-Appraisal-Failure (P-A-F) model. The system enforces mandatory P-A-F classification through a “one-click” stopwatch interface, automatically computes the Appraisal/Failure (A/F) ratio, and delivers real-time strategic recommendations via a WebSocket-powered dashboard. Three user roles are supported: Project Manager, SQA Engineer, and Developer. Seven user requirements were decomposed into 28 system-level requirements spanning five functional features. The system architecture follows a three-layer pattern: a Next.js/React frontend, a Python Flask COQ Engine backend, and a relational database. Development adhered to ISO/IEC 29110 guidelines using Agile methodology. A comprehensive test campaign comprising 68 unit and system test cases achieved a 100% pass rate across all five features, and a complete requirements traceability matrix confirms end-to-end coverage from URS through design artifacts to test validation.

This study presents PetMate, an Animal-Computer-Interaction system for an intelligent cat feeding and health monitoring platform. With the rapid pace of urban life, many cat owners struggle to maintain consistent feeding schedules and monitor their pets' health. Traditional manual feeding is error-prone, while existing smart feeders lack integrated health monitoring. PetMate integrates hardware devices (ESP8266, HX711, DHT11) with cloud services and a web interface, enabling automatic feeding, real-time monitoring, remote control, and health record management. The system supports auto, manual, and timer modes, with weight and diet tracking features. Evaluation results show that 64 unit tests and 65 system tests passed with 100% success rate. The system follows ISO 29110 standards.

Automated unit testing is essential for software quality, yet evaluating the true efficiency of generation tools remains challenging. This paper compares the test generation efficiency of a traditional search-based tool (EvoSuite) and an LLM (GPT-3.5-Turbo) on 15 Java files across varying structural complexities. We evaluate efficiency through statement/branch coverage, test pass rates, and the impact of iterative prompting. Quantitatively, EvoSuite demonstrates superior initial efficiency, achieving high coverage and nearly perfect pass rates, whereas GPT-3.5-Turbo requires multiple iterations to close the structural gap. However, our qualitative analysis of failed assertions reveals that this efficiency comparison masks a “pass rate illusion.” While EvoSuite relies heavily on defensive assertions for runtime stability, iteratively refined GPT-3.5-Turbo tests attempt to verify actual semantic correctness, despite exhibiting more Expected Result Errors. These findings suggest that evaluating test generation efficiency must go beyond traditional coverage and pass rate metrics to include assertion quality.

Excessive screen time among adolescents has become a global concern, leading to cognitive decline, physical inactivity, and psychological issues. Traditional screen-blocking applications often cause frustration and low adherence. This paper presents Screen Time Bank, a novel system that gamifies physical exercise by converting it into a digital currency ("Time Coins") which users spend to unlock entertainment applications. The system integrates real-time pedometer data, customizable exercise plans, and app usage control to motivate healthier habits through immediate rewards. We describe the system architecture, implementation details, and a user scenario demonstrating the "exercise-to-play" reward loop. Preliminary user feedback and usage data indicate increased physical activity and more mindful screen consumption. Future work includes comprehensive user testing and performance evaluation.

The advent of Industry 4.0 has fundamentally transformed traditional manufacturing and industrial operations, shifting the paradigm toward highly automated, intelligent systems. At the core of this transformation lies computer vision, specifically object detection algorithms, which serve as the "eyes" of modern cyber-physical systems. This paper presents an exhaustive, in-depth comparative analysis of the evolutionary trajectory of object detection techniques, charting the development from early two-stage region-based detectors (such as R-CNN and Fast R-CNN) to highly optimized one-stage regression-based architectures (including SSD and YOLOv4), culminating in a profound evaluation of the current state-of-the-art: YOLO11 and its industrial variants (e.g., YOLO-WWBi). Unlike standard reviews that merely aggregate performance metrics, this study synthesizes theoretical architectural shifts with extensive empirical data to uncover the underlying mechanisms that drive detection efficacy. We evaluate these models across stringent parameters including frames per second (FPS), inference latency, parameter efficiency, and mean Average Precision (mAP) under complex real-world constraints. Our original analysis demonstrates that while YOLOv4 successfully democratized real-time processing (achieving 67 FPS), YOLO11 exhibits an absolute paradigm superiority. By integrating mechanisms like Weighted and Re-parameterized Ghost Multi-Scale Feature Aggregation (WRGMSFA) and Bidirectional Feature Pyramid Networks (BiFPN), YOLO11 successfully solves the historical trade-off between semantic granularity and computational overhead. It achieves up to 96.6% mAP in highly complex micro-object tasks (such as Printed Circuit Board defect detection) while maintaining robust real-time reliability on edge devices. Furthermore, this paper proposes novel conceptual frameworks combining YOLO11's edge-computing capabilities with industrial power mitigation strategies, proving that YOLO11 is not merely an algorithmic upgrade, but a foundational catalyst for next-generation sustainable automation.

International students in Thailand face significant barriers in accessing local news, building social connections, and adapting to an unfamiliar environment. This paper presents BridgeU, a bilingual (Chinese/English) web-based community platform designed to address these challenges. BridgeU provides three integrated features: a Daily Briefing system that automatically aggregates Thai news via Google News RSS and translates it into Chinese and English using Qwen AI; a Community Page that enables peer-to-peer knowledge sharing with AI-powered moderation and translation; and an Authentication and Profile system with email and SMS verification. The system is built using Vue.js 3 on the frontend and Spring Boot 3 on the backend, with MySQL 8 as the primary database and JWT-based stateless authentication. A competitive analysis against Facebook Groups, Xiaohongshu, and Reddit confirms that BridgeU uniquely combines bilingual AI news aggregation with a structured, moderated community space. Testing covered 25 unit test suites and 21 system test cases across all features, confirming that all core functional requirements were satisfied.

This study presents "ThaiGo", a mobile educational Role-Playing Game (RPG) designed to bridge the language gap for exchange students from North Minzu University (NMU) studying at Chiang Mai University (CMU). Traditional language learning tools often rely on rote memorization and isolated flashcards, which can lead to social anxiety and ineffective adaptation for new students. To address this, ThaiGo leverages situated learning principles and gamification within a 2D pixel-art simulated campus environment. The system features a core Explore-Learn-Reward loop, dialogue-based learning, and a quest progression system.Developed using Unity 6 for Android, the architecture ensures offline accessibility with JSON-based data persistence. By allowing users to acquire vocabulary through meaningful NPC interactions in realistic scenarios (e.g., ordering food, navigating the campus), this project aims to reduce student anxiety, improve cultural integration, and provide a cost-effective, immersive alternative to premium language platforms.

Automated API testing is crucial for ensuring the re- liability of modern software architectures. While Large Language Models (LLMs) demonstrate significant potential in generating API tests from OpenAPI specifications, existing baseline methods often yield extremely low execution success rates when applied to complex, industrial-grade ecosystems like Java Spring Boot. In this paper, we conduct an empirical study reproducing the RestTSLLM methodology to identify the root causes of these failures. Our findings indicate that these failures primarily stem from the LLM’s lack of environmental context-awareness and its inability to adhere to strict HTTP protocol semantics. To address these limitations, we propose the Constraint-Enhanced Prompting Framework. This approach shifts human intervention from downstream code post-processing to upstream prompt design by introducing three core constraint pillars: Dynamic Context, Protocol Semantics, and Regex Strictness. Evaluated on the widely adopted Spring PetClinic project using the DeepSeek- V3 model, our enhanced framework successfully eliminated generation hallucinations, achieving a 100% execution pass rate in the evaluated test suites. Furthermore, the highly constrained LLM effectively served as a strict API contract auditor, detecting latent real-world vulnerabilities in the System Under Test (SUT), including Specification-to-Implementation drift and unhandled server exceptions.