كلية الامام الكاظم للعلوم الاسلامية الجامعة - Imam Alkadhim University College

This project presents an intelligent academic FAQ system designed to enhance communication between students and university administration by providing accurate and reliable answers to frequently asked questions. The system addresses a common challenge in educational environments, where students ask the same questions in different linguistic forms, dialects, or incomplete expressions, making traditional rule-based or keyword-based systems ineffective. To overcome this limitation, the proposed system adopts a hybrid retrieval approach that combines structured knowledge representation with intelligent matching techniques. Instead of relying on exact text matching, the system analyzes user input through multiple processing stages, including text normalization, keyword extraction, and intent identification. It then compares the processed query against a curated database of canonical questions and their paraphrased variations using scoring metrics such as term overlap, coverage, and anchor word matching. A key innovation of this system is the integration of a safety filtering layer, which ensures that no answer is returned unless the system reaches a sufficient confidence level. This mechanism effectively eliminates the risk of incorrect or misleading responses (commonly known as hallucinations in AI systems), making the system highly reliable for academic use. The system is implemented using Python and Flask for the backend, with a lightweight and scalable database structure for managing FAQs and their associated paraphrases. It supports both formal Arabic and colloquial Iraqi dialects, enabling natural interaction with users without requiring strict input formatting. Additionally, the system includes a dynamic suggestion mechanism that guides users toward valid questions when ambiguity is detected, thereby improving user experience and interaction efficiency. It is designed to be easily extendable, allowing administrators to continuously add new questions and paraphrases without retraining or restructuring the entire system. Overall, the project delivers a practical, scalable, and safe intelligent assistant tailored for academic institutions, with potential applications in universities, educational platforms, and administrative support systems. Its design emphasizes accuracy, reliability, and adaptability, making it a strong candidate for real-world deployment and further development.

Bone tumor diagnosis from X-ray images remains challenging due to subtle radiographic differences between benign and malignant lesions, limited access to specialized radiologists, and severe class imbalance in medical datasets. This research developed and validated a comprehensive three-stage deep learning system for automated bone tumor diagnosis using the publicly available BTXRD dataset (3,746 images). Stage 1 employed EfficientNet-B5 integrated with 16 clinical features (age, gender, anatomical location, projection) for initial triage into normal, benign, or malignant categories. Stage 2 utilized YOLOv8x object detection at 800×800 resolution to localize tumor regions. Stage 3 applied EfficientNetV2-M on YOLO-cropped regions with 23 clinical features to classify tumors into six specific types. We addressed severe class imbalance (137:1 ratio) through intelligent grouping of rare classes, focal loss with class weights, balanced oversampling, and label smoothing (ε=0.05). Results: Stage 1 achieved 94.84% accuracy and AUC 0.989 on 562 test images, with per- class F1-scores of 0.98 (normal), 0.94 (benign), and 0.79 (malignant). Critically, 74.51% malignant recall with zero cases misclassified as normal ensured no cancers were completely missed. Stage 2 optimization improved detection from 51.14% (YOLOv8m@640) to 71.88% [email protected] (YOLOv8x@800), with 79.79% malignant detection. Bounding box detection substantially outperformed segmentation (71.88% vs 11.78%), validating our detection-only approach. Stage 3 achieved 77.94% accuracy and 0.732 macro F1-score on 281 test cases, with YOLO enhancement improving accuracy by +4.2 percentage points over full-image processing. Per-class F1-scores ranged from 0.875 (multiple osteochondromas) to 0.500 (giant cell tumor). Complete processing required 1.2 seconds per case, enabling real-time clinical deployment. This work represents the first comprehensive three-stage system integrating screening, localization, and classification for bone tumors, demonstrating that YOLO-based localization significantly improves classification, detection outperforms segmentation for downstream tasks, and multi-modal fusion enhances diagnostic performance. The system achieved clinically relevant accuracy while handling extreme class imbalance on the largest public bone tumor dataset available.

The global dependence on centralized communication infrastructure renders conventional messaging applications ineffective during network outages, natural disasters, and in geographically remote areas. This paper presents MADA (Mesh Ad-hoc Decentralized Application), a dual-mode decentralized messaging application developed for Android using the Flutter framework and the Dart programming language. MADA operates in two complementary modes: an Offline Mesh Mode, which establishes a Mobile Ad-hoc Network (MANET) through the Google Nearby Connections API by combining Bluetooth Low Energy (BLE) and Wi-Fi Direct to enable fully decentralized peer-to-peer communication without servers, and an Online Hybrid Mode, which utilizes Firebase Authentication, Cloud Firestore, and Firebase Storage for cloud-based message synchronization, media storage, and group management whenever Internet connectivity is available. The system performs seamless automatic failover between both modes, ensuring uninterrupted communication regardless of infrastructure availability. The offline layer employs a store-and-forward routing architecture with end-to-end AES-256-GCM encryption using Elliptic Curve Diffie–Hellman (ECDH) key exchange and multi-hop message relaying. To further improve routing efficiency, MADA integrates an AI-driven heuristic inference engine that performs adaptive multi-signal analysis to determine whether messages should be forwarded, buffered, or dropped, thereby reducing energy consumption, mitigating network congestion, and improving delivery reliability. In addition to text messaging, MADA supports rich media exchange, including images, audio, video, file attachments, real-time GPS location sharing, group conversations, AI-powered smart replies through the Gemini model, and emergency message broadcasting. Experimental evaluation on real Android devices confirmed successful multi-hop routing, sub-second message delivery in both operating modes, and seamless automatic switching between offline and online communication without message loss. The results demonstrate that a unified Flutter-based codebase can provide a robust and adaptive communication platform capable of maintaining reliable messaging in both connected and infrastructure-less environments.

platform that allows citizens and tourists access to Iraq's archaeological heritage. It covers Sumerian, Akkadian, Babylonian, Assyrian, and Islamic sites in Arabic, English, and Kurdish. The app features an innovative interactive system based on GPS, rewarding visitors with badges only for actually visiting sites, not for passive browsing. It also includes an AI-powered chatbot that operates offline, addressing the issue of weak connectivity at many heritage sites. The app also supports content sharing on social media, service reviews, and local data storage, making it lightweight and independent of external servers. It has been tested in 51 scenarios covering its core features and achieved largely successful results, with only minor GPS accuracy issues in areas with weak signals. "Mirath" offers a practical model that combines heritage content, site verification, and offline AI, paving the way for future collaboration with cultural and tourism authorities.

يقوم مشروعنا على حفظ سلامة رجال الاطفاء من الغازات السامة و نقص الأوكسجين وذلك عبر إرسال المعلومات اللازمه إلى مركز التحكم الموجود في مركز الإطفاء وذلك بتحديد موقع رجل الاطفاء من خلال GPS و اخطارات الغازات من خلال قراءات الحساسات وذلك عبر الخوذة الذكية التي يرتديها رجل الاطفاء وباستخدام انترنت الاشياء و الذكاء الاصطناعي بحسب المنطق المضببfuzzy logic.

The rapid evolution of labor market requirements has created a significant gap between academic education and industry expectations. Many students and graduates face difficulties in identifying the skills required for their target careers and evaluating their readiness for employment. This project presents a Smart Skill Recommendation System Based on Labor Market Needs that utilizes Artificial Intelligence, Natural Language Processing (NLP) and Machine Learning techniques to analyze academic and job-related data. The proposed system extracts skills from academic courses and job descriptions, compares them using TF-IDF and Cosine Similarity algorithms, and identifies missing competencies through Skill Gap Analysis. The platform provides personalized recommendations, compatibility scores, visual analytics, and career guidance to help users improve their employability. The expected outcomes include enhancing career planning, reducing the gap between education and employment and supporting students in making informed decisions regarding skill development and professional growth. Keywords: Artificial Intelligence, NLP, Skill Gap Analysis, Career Recommendation, Machine Learning.