# **Final Year Project Proposal & SRS**
**Title**: **VeriGuard: A Machine Learning–Based Fake News Detection System**
## **1. Introduction**
### **1.1 Background**
In the digital era, information spreads at unprecedented speed through social media, messaging apps, and online news platforms. Unfortunately, so does misinformation. Fake news—deliberately fabricated or misleading content—can influence public opinion, incite panic, and damage reputations. Studies show that fake news spreads significantly faster and farther than true stories, often due to its emotionally charged nature.
Existing fact-checking platforms (e.g., Snopes, FactCheck.org) rely heavily on manual verification, which is slow and cannot scale to real-time demands. There is a critical need for an automated, intelligent system that can analyze textual content and flag potentially false information using machine learning (ML).
### **1.2 Problem Statement**
Fake news spreads faster than factual reporting, especially on social media. Manual verification is time-consuming and cannot keep pace with the volume of content generated daily. Users lack immediate, reliable tools to assess news credibility before sharing or acting upon it.
### **1.3 Proposed Solution**
**VeriGuard** is an AI-powered web and mobile application that uses natural language processing (NLP) and supervised machine learning models to detect fake news in real time. The system analyzes article content, linguistic patterns, source credibility, and metadata to classify news as **“Likely Real,” “Likely Fake,”** or **“Uncertain.”** It provides users with a credibility score, supporting evidence, and alternative verified sources.
---
## **2. Objectives**
### **2.1 General Objective**
To design, develop, and evaluate a scalable machine learning system that accurately detects fake news in textual content and educates users on information credibility.
### **2.2 Specific Objectives**
1. Collect and preprocess a diverse dataset of real and fake news articles.
2. Train and evaluate multiple ML/NLP models (e.g., BERT, LSTM, SVM, Random Forest) for fake news classification.
3. Develop a responsive web application with a RESTful backend and intuitive UI.
4. Implement real-time URL/article analysis with explainable AI (XAI) features.
5. Integrate user feedback to enable continuous model improvement.
6. Evaluate system performance using accuracy, precision, recall, and F1-score.
7. Ensure data privacy, low latency, and cross-platform accessibility.
---
## **3. Scope**
### **3.1 In Scope**
- Text-based fake news detection (English language only in v1.0).
- Web application accessible via desktop and mobile browsers.
- URL input and direct text paste functionality.
- Real-time classification with confidence score (0–100%).
- Source reputation database (curated list of trusted/untrusted domains).
- User feedback mechanism (thumbs up/down).
- Admin dashboard for model monitoring and dataset management.
- RESTful API for third-party integration (e.g., browser extensions).
### **3.2 Out of Scope**
- Image, video, or audio-based misinformation detection.
- Multilingual support (beyond English).
- Social media account verification.
- Legal enforcement or content takedown.
- Real-time social media feed monitoring (e.g., Twitter/X scraping).
---
## **4. Functional Requirements**
| **ID** | **Feature** | **Description** |
|--------|-------------|------------------|
| FR-01 | **User Registration/Login** | Users can create accounts using email or Google SSO. |
| FR-02 | **News Submission** | Users can submit news via URL or paste article text. |
| FR-03 | **Real-Time Analysis** | System processes input and returns classification within 3 seconds. |
| FR-04 | **Credibility Report** | Displays: classification label, confidence score, key indicators (e.g., sensational language, unverified claims), and links to verified sources. |
| FR-05 | **Source Reputation Lookup** | Checks domain against a curated trustworthiness database. |
| FR-06 | **User Feedback** | Users can rate prediction accuracy; feedback stored for model retraining. |
| FR-07 | **History Log** | Users can view past analyses with timestamps and results. |
| FR-08 | **Admin Dashboard** | Admins can view system metrics, retrain models, and manage source trust list. |
| FR-09 | **API Access** | Developers can integrate fake news detection via `/analyze` endpoint. |
| FR-10 | **Explainability Panel** | Highlights suspicious phrases and explains why they triggered fake flags (e.g., “This sentence contains emotionally charged language”). |
---
## **5. Non-Functional Requirements**
| **Category** | **Requirement** |
|-------------|------------------|
| **Performance** | Response time ≤ 3 seconds for 95% of requests under 100 concurrent users. |
| **Accuracy** | ≥ 92% F1-score on benchmark datasets (e.g., FakeNewsNet, LIAR). |
| **Usability** | Intuitive UI; usable by non-technical users; WCAG 2.1 AA compliant. |
| **Security** | HTTPS, input sanitization, rate limiting, GDPR-compliant data handling. |
| **Reliability** | 99.5% uptime; auto-recovery from failures. |
| **Scalability** | Support 1,000+ daily users; model inference via containerized microservices. |
| **Maintainability** | Modular codebase (Python/Django + React); CI/CD pipeline. |
| **Privacy** | No permanent storage of user-submitted text; analytics anonymized. |
---
## **6. System Architecture**
### **6.1 High-Level Design**
- **Frontend**: React.js (responsive PWA)
- **Backend**: Django REST Framework (Python)
- **ML Engine**: Scikit-learn + TensorFlow/PyTorch (hosted as microservice)
- **Database**: PostgreSQL (user data), Redis (caching)
- **Model Serving**: Flask API wrapped in Docker container
- **Deployment**: AWS EC2 or Heroku (with CI/CD via GitHub Actions)
### **6.2 Data Flow**
1. User submits URL/text → frontend → backend API.
2. Backend fetches article content (if URL) → cleans text → sends to ML service.
3. ML service runs NLP pipeline → returns prediction + explanation.
4. Result displayed to user + stored in history (anonymized).
5. Feedback loop: user ratings trigger model retraining weekly.
---
## **7. Machine Learning Approach**
### **7.1 Dataset**
- **Primary**: FakeNewsNet (politifact & gossipcop subsets)
- **Supplementary**: LIAR dataset, ISOT Fake News Dataset
- **Preprocessing**: Lowercasing, stopword removal, lemmatization, TF-IDF/BERT embeddings
### **7.2 Models Evaluated**
- Logistic Regression (baseline)
- SVM with TF-IDF
- LSTM with GloVe embeddings
- **Fine-tuned DistilBERT** (selected for balance of speed/accuracy)
### **7.3 Evaluation Metrics**
- Accuracy, Precision, Recall, F1-Score
- Confusion Matrix, ROC-AUC
- Cross-validation (5-fold)
---
## **8. Development Methodology**
- **Agile (Scrum)**: 2-week sprints, backlog grooming, daily standups (simulated)
- **Tools**: Jira (task tracking), GitHub (version control), Figma (UI design)
- **Testing**: Unit tests (pytest), E2E (Cypress), model validation (MLflow)
---
## **9. Expected Outcomes**
1. A fully functional web application for fake news detection.
2. A trained ML model with ≥92% F1-score on benchmark data.
3. Technical documentation, user manual, and source code repository.
4. Dissertation report meeting University of Moratuwa BIT standards.
---
## **10. Timeline (Gantt Overview)**
| **Phase** | **Duration** | **Deliverables** |
|----------|-------------|------------------|
| Requirement Analysis | Weeks 1–2 | SRS, Use Cases |
| System Design | Weeks 3–4 | Architecture, DB Schema, UI Mockups |
| Dataset Curation & Model Training | Weeks 5–8 | Trained Model, Evaluation Report |
| Frontend & Backend Dev | Weeks 9–12 | MVP Application |
| Integration & Testing | Weeks 13–14 | Test Reports, Bug Fixes |
| Deployment & Documentation | Weeks 15–16 | Live Demo, Final Report |
---
## **11. References**
1. Shu, K., et al. (2017). *Fake News Detection on Social Media: A Data Mining Perspective*. ACM SIGKDD.
2. Wang, W. Y. (2017). *“Liar, Liar Pants on Fire”: A New Benchmark Dataset for Fake News Detection*. ACL.
3. TensorFlow & Hugging Face Documentation.
4. IEEE SRS Template (IEEE Std 830-1998).
---
This document is **dissertation-ready** and includes sufficient technical depth for implementation. You may expand sections like **risk analysis**, **ethical considerations**, or **detailed UML diagrams** (use case, sequence, ERD) as needed for your final submission.
Would you like the **database schema**, **API specification**, or **UI wireframes** next?
💥 WordPress https://computerclassinsrilanka.wordpress.com
💥 Facebook https://web.facebook.com/itclasssrilanka
💥 Wix https://itclasssl.wixsite.com/icttraining
💥 Web https://itclasssl.github.io/eTeacher/
💥 Medium https://medium.com/@itclasssl
💥 Quora https://www.quora.com/profile/BIT-UCSC-UoM-Final-Year-Student-Project-Guide
💥 mystrikingly https://bit-ucsc-uom-final-year-project-ideas-help-guide-php-class.mystrikingly.com/
💥 https://elakiri.com/threads/bit-ucsc-uom-php-mysql-project-guidance-and-individual-classes-in-colombo.1627048/
💥 https://bitbscucscuomfinalprojectclasslk.weebly.com/
💥 https://www.tiktok.com/@onlinelearningitclassso1
💥 https://payhip.com/eTeacherAmithafz/
💥 https://discord.gg/cPWAANKt
💥 https://6789f6ca695da.site123.me/blog
💥 https://itclasssl.github.io/eTeacher/
💥 https://www.tumblr.com/blog/computercoursesincolombo-blog
💥Subscribe on LinkedIn https://www.linkedin.com/build-relation/newsletter-follow?entityUrn=7072056309516304384
💥https://www.scribd.com/user/682883198/Online-Learning-IT-Class-Software-Dev
🚀 English Sinhala Tamil Medium Online Classes
✅ Final Year Software Web Project Guidance BIT UCSC UoM
✅ Grade 1 to GCE O/L A/L ICT GIT Classes
✅ PHP & Python Training
✅ Web & Software Development
✅ Social Media Marketing
📲 Learn, Build & Succeed! Join us today! 🚀
🚀 Looking for HIGH-ENGAGEMENT Student Groups to Share Your IT/CS Project Ideas?
Here’s the ULTIMATE list to get MASSIVE reach in 2025! 💥💥💥
🧨 1. University IT/CS Students Groups (SUPER ACTIVE!)
Search Keywords:
Computer Science Students • IT Students Community • Software Engineering Students • BSc IT Students • Final Year Project Group
🌏 2. Indian Engineering Groups (HUGE REACH!)
Millions of BTech/BCA/MCA students!
Search:
BTech Projects • Engineering Students India • BCA MCA Students • CSE Students Group • Polytechnic IT Group
🇧🇩 3. Bangladesh CS/IT Groups (VERY ENGAGED!)
Search:
CSE Bangladesh • ICT Bangladesh • University Students BD • Final Year Project BD
🇵🇰 4. Pakistan IT & CS Groups
Search:
CS Pakistan • IT Students Pakistan • Final Year Projects Pakistan • BSCS Students Pakistan
🇱🇰 5. Sri Lankan IT/SE Groups (HOME ADVANTAGE!)
Search:
UCSC Groups • SLIIT Groups • IT Students Sri Lanka • BIT External Groups • SL IT Jobs & Projects
💻 6. Global Programming & Coding Groups
Search:
Python Projects • JavaScript Developers • Web Developers Community • MERN Stack Devs • Full-Stack Developers
🤖 7. AI & Machine Learning Groups
Search:
AI Projects • Machine Learning Community • Data Science Projects • AI Engineers Group
📱 8. App Development Groups
Search:
Android Project Ideas • Flutter Developers • Mobile App Developers
🧑🎓 9. Assignment & Academic Help Groups
Search:
Assignment Help • University Students Help • Homework Help • Academic Projects
🌍 10. Tech Learning & Career Groups
Search:
Tech Learners Community • Learn Programming • Coding For Beginners • Computer Science Hub
🎯 Pro Tips for MAXIMUM Reach
🔥 Join 30–40 groups
🔥 Post 5–6 times per day
🔥 Change your caption slightly each time
🔥 Use strong hooks like:
“🔥 Final Year IT Project Idea (Problem + Solution)”
“💡 Real-World IT Problem You Can Solve With AI!”
“🚀 Best Project for BSc/BIT/MCA Students!”
#️⃣ Hashtags to BOOST Reach
#ITProjects #CSStudents #FinalYearProject #SoftwareEngineering #ComputerScience #BTechStudents #UniversityProjects #ProgrammingIdeas #AIProjects #WebDevelopment #MobileDevelopment #CodingCommunity #SriLankaIT #IndiaEngineering #BangladeshCSE #PakistaniStudents #StudentProjects #ProjectIdeas2025
