README.md

🎓 HelpMate - AI-Based Personalized Learning & Wellness Assistant

Transforming Academic Experience Through Intelligent Automation

Project ID: 25-26J-228
Research Group: CoEAI - Centre of Excellence for AI
Specialization: Information Technology

---

📋 Table of Contents

• Overview
• The Problem
• The Solution
• System Modules
  • Module 1: Intelligent Study Content Generation
  • Module 2: Smart Academic Group Work Management
  • Module 3: Adaptive Student Task Scheduler
  • Module 4: Community & Collaboration Zone
• System Architecture
• Tech Stack
• Innovation Highlights
• Installation & Setup
• Research Objectives
• Team
• Future Enhancements
• References

---

🌟 Overview

HelpMate is an all-in-one AI-powered platform designed to revolutionize the academic experience for undergraduate students in Sri Lankan universities. It integrates intelligent academic planning, wellness guidance, collaborative learning, and performance analytics into a unified, adaptive ecosystem.

🎯 What Makes HelpMate Special?

An integrated, AI-driven solution that merges academic personalization with emotional intelligence and peer-to-peer interaction in a responsive environment.

HelpMate addresses the fragmented nature of current educational tools by providing:

• ✅ Personalized Study Planning with AI-driven content generation
• ✅ Fair Group Work Management with contribution tracking
• ✅ Intelligent Task Scheduling using Reinforcement Learning
• ✅ Peer Collaboration through ML-based matching
• ✅ Emotional Wellness Support with mood tracking
• ✅ Career Path Guidance with CV analysis and job matching

---

🔴 The Problem

Challenges Faced by Students Today

📚 Academic Struggles Overwhelming volume of content without structured tools Difficulty converting notes into active learning Poor time management and procrastination Inability to identify weak topics early Last-minute cramming and exam anxiety	👥 Collaboration Issues Free-riding in group projects Lack of transparency in contribution tracking Unfair academic evaluation Poor coordination among team members Difficulty finding compatible study partners
⏰ Planning & Productivity Static to-do lists that don't adapt No consideration of energy patterns Inefficient task prioritization Ignoring fatigue and subject difficulty One-size-fits-all scheduling approaches	🎯 Career & Community Fragmented digital resources Lack of mentorship networks No career path guidance Isolated learning experiences Missing skill gap identification

🎓 Research Gap

Traditional learning platforms focus solely on content delivery, with no regard for individual needs, emotional factors, or collaborative dynamics. Existing tools operate in isolation, forcing students to juggle multiple disconnected applications without adaptive intelligence.

---

✨ The Solution

HelpMate: Four Integrated AI Modules

HelpMate brings together four fundamental components under one intelligent roof:

graph TD
    A[Student Profile & Behavior] --> B[Smart Learning Profiler]
    B --> C[Personalized Study Planner]
    C --> D[Content Generation Module]
    D --> E[RL Task Scheduler]
    E --> F[Group Work Manager]
    F --> G[Community & Collaboration]
    G --> H[Performance Analytics]
    H --> A

Loading

---

🧩 System Modules

1️⃣ Intelligent Study Content Generation Module

🎯 Purpose

Automatically transforms uploaded lecture materials into interactive learning assets—quizzes, flashcards, and summarized notes—while intelligently aligning delivery with exam deadlines.

✨ Key Features

📄 Lecture Note Upload & Processing

• Upload PDFs via drag-and-drop interface
• Secure storage using Multer
• Text extraction with pdf-parse
• Automatic cleaning and segmentation

🤖 AI-Powered Generation (Fine-tuned T5 Model)

📝 MCQ Generation Q: What is normalization? A) Data encryption B) Reducing redundancy ✓ C) Query optimization D) Index creation

🗂️ Flashcards Front: ACID Properties Back: Atomicity, Consistency, Isolation, Durability

📄 Summaries Abstractive summary using BART/T5: "Database normalization eliminates redundancy through normal forms..."

📅 Deadline-Based Adaptive Delivery

Time Until Exam	Strategy	Frequency
> 3 weeks	Generate all content	Weekly reviews
2–3 weeks	Spaced repetition	Every 5 days
1–2 weeks	Focus weak areas	Every 2-3 days
< 1 week	Intensive revision	Daily drills

🎯 Interactive Quiz Engine

• Multiple-choice interface with instant feedback
• Score tracking and weak topic detection
• Performance data feeds into prediction models

🔬 Technical Implementation

// Content Generation Pipeline
Upload PDF → Extract Text → NLP Processing (T5/BERT) 
→ Generate MCQs/Flashcards/Summaries → Store → Schedule Delivery
→ Track Performance → Feed Prediction Engine

Tech Stack:

• NLP Models: Fine-tuned T5-base, BERT, BART
• Backend: Node.js, Express, Python Flask
• Processing: spaCy, Hugging Face Transformers

---

2️⃣ Smart Academic Group Work Management System

🎯 Purpose

Improve fairness and transparency in university group projects by tracking individual contributions, reducing free-riding, and supporting better academic evaluation.

✨ Key Features

📝 Project & Task Management

• Create and manage group projects
• Assign tasks to members with deadlines
• Track completion status in real-time
• Group leader dashboard for coordination
• Student self-management tools

📊 Contribution & Performance Tracking

Metrics Tracked:
├── Number of tasks completed
├── Contribution percentage
├── Active time (project interactions)
├── Task complexity weighting
└── Participation frequency

🔍 Free-Riding Detection (Rule-Based)

Identifies low participation using:

• Active time significantly below team average
• Task completion count outliers
• Contribution percentage thresholds
• Automated alerts for supervisors

📈 Analytics Dashboard

Quick Overview Cards 📊 Total open projects ✅ To-do tasks 🔄 Ongoing tasks 📅 Upcoming deadlines

Visual Insights 🥯 Donut charts for task distribution 📊 Line charts for daily activity 📈 Project-wise activity filters ⏰ Last active timestamps

🤖 ML-Based Contribution Scoring

# Research Component: Predictive Scoring
Model: Random Forest / Gradient Boosting
Features: [active_time, tasks_completed, complexity_score, 
          interaction_frequency, commit_history]
Output: Contribution Score (0-100)
Training: FastAPI + Google Colab

• Training separate from main system
• Predictions used for scoring only (not decision-making)
• Frontend displays predicted contribution scores
• Helps supervisors make informed evaluations

🔬 Technical Implementation

Backend: Node.js, Express
ML Server: Python FastAPI
Visualization: Chart.js, D3.js
Database: Firebase Firestore

---

3️⃣ Adaptive Student Task Scheduler with Reinforcement Learning

🎯 Purpose

An intelligent, personalized study planner that fights procrastination by learning your energy patterns using Reinforcement Learning.

💡 The Innovation

Traditional to-do lists are static. They don't care if you're tired, overwhelmed, or biased against Math.

This Next-Generation Scheduler uses RL to dynamically generate daily study plans by learning:

• 🌅 Are you a morning person or night owl?
• ⏱️ Do you underestimate coding assignment durations?
• 😴 Are you too fatigued for a Difficulty-5 Physics task right now?

✨ Key Features

🤖 PPO Reinforcement Learning Agent

State Space:
├── Current fatigue level (0-10)
├── Time of day (0-23)
├── Subject difficulty (1-5)
├── Recent focus ratings
├── Deadline urgency
└── Historical completion rates

Action Space:
├── Schedule task in time slot
├── Skip to next slot
└── Mark break time

Reward Function:
R = α(task_completion) + β(focus_quality) 
    - γ(deadline_penalty) - δ(fatigue_violation)

Training Process:

• Tracks Pomodoro sessions (25-min focus blocks)
• Records focus ratings after each session
• Learns subject-level strengths/weaknesses
• Adapts to energy patterns over time

📊 The Pepsi Challenge: AI vs. Heuristic

Method	Approach	Advantages	Limitations
🎯 Heuristic Baseline	Priority + Deadline Greedy Algorithm	• Fast execution • Predictable results • Easy to understand	• Rigid ordering • Ignores energy levels • No personalization
🤖 RL Agent	PPO-based Dynamic Scheduling	• Adapts to fatigue • Learns preferences • Optimizes long-term productivity	• Requires training data • Computationally intensive • Black-box decisions

📡 API Endpoints for Testing

# Get Heuristic Schedule (Baseline)
GET /api/schedule/heuristic
# Result: Strict deadline ordering - efficient but rigid

# Get AI Schedule (RL Agent)
GET /api/schedule/rl
# Result: Dynamic ordering by priority + energy patterns

⏰ Smart Task Management

• Tasks divided into 25-minute Pomodoro sessions
• In-progress tasks stay prioritized (momentum maintenance)
• Exam-linked tasks get higher urgency
• Respects sleep schedules and class timings
• Avoids reward hacking through balanced objectives

🔬 Technical Implementation

RL Framework: Stable-Baselines3 (PPO)
Backend: Node.js, Express
Training: Python, TensorFlow/PyTorch
State Management: Redis for fast lookups
Deployment: Separate microservice architecture

---

4️⃣ Community & Collaboration Zone

🎯 Purpose

Facilitate peer collaboration, mentorship networks, and career development through intelligent matching algorithms and CV analysis.

✨ Key Features

🤝 Peer Matching via K-Means Clustering

Mathematical Foundation:

The algorithm minimizes Within-Cluster Sum of Squares (WCSS):

WCSS = Σ(i=1 to k) Σ(x ∈ Ci) ||x - μi||²

where:
- k = number of clusters
- Ci = cluster i
- μi = centroid of cluster i
- x = student profile vector

Implementation Process:

1. Data Vectorization: Convert student profiles (interests, skills, goals) into numerical vectors
2. Clustering: Use ml-kmeans library to partition students into k groups
3. Matching: Assign students to clusters with nearest centroid
4. Refinement: Self-improving algorithm with reinforcement learning

Features Considered:

• Academic interests and specialization
• Programming language proficiency
• Project experience and skills
• Learning style preferences
• Availability and timezone

🎓 Mentorship Network

• Senior-Junior Pairing: Connect experienced students with newcomers
• Adaptive Pairing: Dynamic adjustment based on interaction quality
• Academic Guidance: Domain expertise sharing
• Skill Development: Collaborative learning opportunities
• Compatibility Ratings: Feedback-driven refinement

📄 CV Analysis & Career Path Suggestions

PDF Extraction Pipeline:

Upload CV (PDF) → Extract Text (pdf-parse / pdfjs-dist)
→ Pattern Matching (Regex + NLP) → Identify Skills/Experience
→ Compare with Industry Standards → Generate Skill Gap Report
→ Suggest Certifications/Projects → Job API Integration

Components:

1️⃣ Text Extraction

const pdfParse = require('pdf-parse');
// Extract raw text streams from PDF
const data = await pdfParse(cvBuffer);
const cvText = data.text;

2️⃣ Pattern Matching & NLP

// Extract key sections using Regex
const skills = extractSection(cvText, /skills?:/i);
const experience = extractSection(cvText, /experience:/i);
const education = extractSection(cvText, /education:/i);

3️⃣ Skill Gap Analysis

// Compare against industry requirements
const requiredSkills = getCareerRequirements(targetRole);
const missingSkills = requiredSkills.filter(
  skill => !skills.includes(skill)
);

4️⃣ Job API Integration

// Asynchronous HTTPS requests to job platforms
const jobResults = await axios.get('https://api.linkedin.com/jobs', {
  params: { skills: extractedSkills, location: 'Sri Lanka' }
});

Supported APIs:

• 🔗 LinkedIn Jobs API
• 🌐 Indeed API
• 🎯 Adzuna API
• 📊 Glassdoor API

📊 Career Analytics Engine

• Personalized Recommendations: Based on current profile
• Skill Gap Identification: What's missing for target roles
• Market Demand Analysis: Trending skills in job market
• Certification Suggestions: Relevant courses and certifications
• Predictive Modeling: Career success probability estimation
• Learning Path Generation: Step-by-step skill development plan

🎯 Goal-Sharing Dashboards

• Collaborative goal tracking among peers
• Progress visualization and milestones
• Motivational analytics and achievements
• Study group formation tools
• Project collaboration features

🔬 Technical Implementation

Clustering: ml-kmeans (Node.js ML library)
PDF Processing: pdf-parse, pdfjs-dist
NLP: Basic pattern matching + Regex
Job APIs: axios for HTTP requests
File Upload: multer middleware
Backend: Node.js, Express

📦 Key Dependencies

Library	Purpose
express	Web framework for API routing
pdf-parse	Extracting text from CVs
ml-kmeans	Clustering logic for peer matching
axios	API calls to job platforms
multer	Handling PDF uploads
natural	NLP utilities
compromise	Text parsing

---

🏗️ System Architecture

Overall Platform Architecture

Component Flow Diagram

┌─────────────────────────────────────────────────────────────┐
│                    React Frontend (Port 3000)                │
│  ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐      │
│  │ Content  │ │  Group   │ │    RL    │ │Community │      │
│  │   Gen    │ │   Work   │ │Scheduler │ │   Zone   │      │
│  └──────────┘ └──────────┘ └──────────┘ └──────────┘      │
└────────────────────┬────────────────────────────────────────┘
                     │
          ┌──────────┴──────────┐
          │                     │
┌─────────▼─────────┐  ┌────────▼────────┐
│  Node.js Backend  │  │  Python AI      │
│   (Express API)   │◄─┤  Servers        │
│   Port 8080       │  │  Flask/FastAPI  │
└─────────┬─────────┘  │  Port 4000      │
          │            └─────────────────┘
          │
    ┌─────┴─────┬──────────┬───────────┐
    │           │          │           │
┌───▼───┐ ┌────▼────┐ ┌───▼────┐ ┌───▼────┐
│Firebase│ │ Google  │ │  Redis │ │  Job   │
│  Store │ │Calendar │ │ Cache  │ │  APIs  │
└────────┘ └─────────┘ └────────┘ └────────┘

Microservices Architecture

Service	Technology	Port	Purpose
🎨 Frontend	React, HTML5, CSS3	3000	User interface
⚙️ Main API	Node.js, Express	8080	Core business logic
🤖 Content Gen AI	Python, Flask, T5	4000	NLP & content generation
📊 ML Scoring	Python, FastAPI	4001	Contribution predictions
🧠 RL Scheduler	Python, TensorFlow	4002	Task scheduling agent
🗄️ Database	Firebase Firestore	-	Data persistence
📅 Calendar	Google Calendar API	-	Exam scheduling
💾 Cache	Redis	6379	Fast state lookups

---

💻 Tech Stack

Frontend Technologies

Backend Technologies

AI & Machine Learning

Database & Storage

Complete Technology Matrix

Category	Technologies
NLP Models	T5-base (fine-tuned), BERT, BART, spaCy
ML Algorithms	K-Means Clustering, Random Forest, PPO (RL)
Libraries	Transformers, Stable-Baselines3, ml-kmeans, pdf-parse
APIs	Google Calendar, LinkedIn Jobs, Indeed, Adzuna
DevOps	Docker, GitHub Actions, Heroku/AWS

---

💡 Innovation Highlights

🌟 Groundbreaking Features

Module	Innovation	Impact
📝 Content Generation	Fine-tuned T5 on educational content	First local system with exam-aware scheduling
👥 Group Management	ML-based free-rider detection	Fair evaluation through objective metrics
🧠 RL Scheduler	PPO agent learns energy patterns	Personalized planning beats static to-do lists
🤝 Peer Matching	K-Means clustering on profile vectors	Intelligent collaboration > random grouping
📄 CV Analysis	Automated skill gap identification	Career guidance with job market integration
🔄 Closed-Loop System	All modules feed each other	Holistic improvement > isolated tools

🚀 Competitive Advantages

vs. Traditional Tools ✅ Fully automated vs. manual input ✅ Adaptive vs. static scheduling ✅ Integrated vs. fragmented apps ✅ AI-driven vs. rule-based ✅ Predictive vs. reactive

vs. Generic AI Tools ✅ Domain-specific training vs. prompts ✅ Behavioral learning vs. one-shot ✅ Multi-agent ecosystem vs. single tool ✅ Sri Lankan context vs. global generic ✅ Educational psychology vs. pure tech

🎯 Research Contributions

1. Novel RL Application in academic task scheduling with fatigue modeling
2. Hybrid Evaluation framework (The Pepsi Challenge: AI vs. Heuristic)
3. Multi-dimensional clustering for peer matching beyond simple attributes
4. Closed-loop learning where all modules enhance each other
5. Context-aware content delivery based on deadline proximity
6. Fair assessment methodology for group work using ML scoring

Unlike Quizlet, Anki, Trello, or ChatGPT wrappers, HelpMate is a purpose-built AI ecosystem designed from the ground up for Sri Lankan undergraduate students.

---

🛠️ Installation & Setup

Prerequisites

✅ Node.js (v14+)
✅ Python (v3.8+)
✅ npm or yarn
✅ Git
✅ Redis (optional, for caching)

📦 Clone Repository

git clone https://github.com/osandalakshitha/helpmate.git
cd helpmate

🔧 Backend Setup

cd backend
npm install

# Install dependencies
npm install express multer pdf-parse axios ml-kmeans firebase-admin

npm start  # Runs on port 8080

🤖 AI Servers Setup

Content Generation Server (Port 4000)

cd ai_servers/content_generation
pip install -r requirements.txt
python app.py

requirements.txt:

fastapi
uvicorn
scikit-learn
pandas
numpy
joblib

RL Scheduler Server (Port 4002)

cd ai_servers/rl_scheduler
pip install -r requirements.txt
python scheduler_service.py

requirements.txt:

stable-baselines3
tensorflow
gym
redis
numpy

🎨 Frontend Setup

cd frontend
npm install

# Install additional dependencies
npm install axios chart.js react-router-dom

npm start  # Runs on port 3000

🔑 Environment Variables

Create .env files in respective directories:

Backend (.env):

PORT=8080
NODE_ENV=development

# Firebase
FIREBASE_API_KEY=your_firebase_key
FIREBASE_PROJECT_ID=your_project_id

# APIs
GOOGLE_CALENDAR_API_KEY=your_calendar_key
LINKEDIN_API_KEY=your_linkedin_key
INDEED_API_KEY=your_indeed_key

# AI Servers
CONTENT_GEN_URL=http://localhost:4000
ML_SCORING_URL=http://localhost:4001
RL_SCHEDULER_URL=http://localhost:4002

# Redis
REDIS_HOST=localhost
REDIS_PORT=6379

AI Servers (.env):

FLASK_PORT=4000
MODEL_PATH=./models/t5-base-finetuned
DEVICE=cuda  # or 'cpu'

🚀 Quick Start with Docker (Optional)

# Build all services
docker-compose up --build

# Or run individually
docker-compose up frontend
docker-compose up backend
docker-compose up ai-services

---

📖 Usage Guide

Module 1: Content Generation

1. Upload Notes: Navigate to /upload and drag-drop PDFs
2. Set Exam Date: Input deadline in dashboard
3. Review Content: Access generated MCQs, flashcards, summaries
4. Take Quizzes: Complete assessments and track scores

Module 2: Group Work Management

1. Create Project: Set up group project with members
2. Assign Tasks: Distribute work with deadlines
3. Track Activity: Monitor contribution in real-time
4. View Analytics: Check individual performance metrics
5. Generate Reports: Export contribution summaries

Module 3: RL Task Scheduler

1. Add Tasks: Input assignments with deadlines and difficulty
2. Get Schedule: Compare AI vs. Heuristic plans
3. Complete Pomodoros: Work in 25-min blocks
4. Rate Focus: Provide feedback after sessions
5. Agent Learns: System adapts to your patterns

Module 4: Community & Career

1. Upload CV: Submit resume for analysis
2. View Matches: See compatible study partners
3. Skill Gaps: Review missing competencies
4. Job Suggestions: Browse relevant opportunities
5. Connect: Join study groups or find mentors

---

🎯 Research Objectives

Main Objective

To develop HelpMate — an integrated, AI-based platform that combines personalized study planning, fair group work management, intelligent task scheduling, and peer collaboration to enhance academic outcomes and emotional well-being for undergraduate students.

Module-Specific Objectives

Member	Module	Sub-Objectives	Novelty
S. N. Ilukwaththage IT22609908	Smart Learning Profiler	• Classify learning styles using ML • Generate dynamic study plans • Predict academic performance • Integrate behavior tracking	AI-driven learning style identification with behavior-based performance prediction
S.M.O.L. Chamikara IT22596734	Content Generation & Productivity	• Implement NLP-based flashcard/MCQ generator • Develop smart task manager with reminders • Enable short note generation • Track progress visually	Combines personalized content generation with interactive productivity tools and deadline-aware delivery
R.A.D.B. Vishmi IT22926326	Wellness & Emotion Assistant	• Build mood tracking interface • Integrate emotion detection (NLP) • Design habit tracker • Create relaxation tools	Adds mood-awareness to academic tools using sentiment-based journaling insights
M.M. Sandeep IT22221100	Community & Collaboration	• Implement smart peer matching (K-Means) • Build mentorship network • Create career analytics engine • Develop CV analysis system	Self-improving algorithm with RL for peer matching + predictive career modeling with skill gap analysis

---

📅 Project Timeline

Week	Phase	Deliverables
1-2	📋 Planning	Requirements gathering, feasibility study, dataset collection
3-4	🎨 Design	System architecture, UI/UX wireframes, API specifications, database schema
5-6	🤖 AI Development	T5 fine-tuning, K-Means implementation, RL agent training
7-8	💻 Implementation	Module development, API integration, frontend components
9-10	🔗 Integration	Multi-agent communication, calendar sync, testing workflows
11-12	🧪 Testing & Deployment	User testing, performance optimization, documentation, launch

Methodology: Agile-Scrum

• 2-week sprints with sprint reviews
• Daily standups for team coordination
• Peer code reviews before merging
• Continuous integration with GitHub Actions

---

👥 Team

🎓 HelpMate Research Group - SLIIT

S. N. Ilukwaththage IT22609908 🧠 Smart Learning Profiler & Performance Prediction 📧 Email

S.M.O.L. Chamikara IT22596734 📝 Intelligent Content Generation & Resource Management 📧 Email

R.A.D.B. Vishmi IT22926326 💚 Academic Wellness & Emotion Assistant 📧 Email

M.M. Sandeep IT22221100 🤝 Community & Collaboration Zone with Career Analytics 📧 Email

👩‍🏫 Supervision Team

Supervisor: Mrs. Uthpala Samarakoon
Co-Supervisor: Ms. Tharushi Rubasinghe

---

Institution: Sri Lanka Institute of Information Technology (SLIIT)
Faculty: Faculty of Computing
Research Group: CoEAI - Centre of Excellence for AI
Year: 2025

---

🔮 Future Enhancements

Phase 1: Enhanced Intelligence (Q1 2026)

Feature	Description	Priority
🔍 OCR Support	Extract text from scanned PDFs using Tesseract.js	High
🎙️ Voice Notes	Audio lecture transcription and summarization	High
📊 Advanced Analytics	Predictive GPA calculator based on current performance	High
🌐 Multi-language	Support for Sinhala and Tamil content	Medium

Phase 2: Collaboration 2.0 (Q2 2026)

Feature	Description	Priority
🎮 Gamification	XP, badges, leaderboards for motivation	High
📱 Mobile Apps	Native iOS/Android applications	High
🤝 Live Study Rooms	Virtual co-working spaces with video	Medium
💬 Real-time Chat	Integrated messaging for study groups	Medium
🏆 Competitions	Quiz tournaments and coding challenges	Low

Phase 3: Advanced AI (Q3 2026)

Feature	Description	Priority
🧠 GPT Integration	Advanced question answering on notes	High
🎯 Personalized Tutoring	AI tutor for concept clarification	High
📈 Predictive Interventions	Early warning system for at-risk students	High
🔮 Career Path Simulator	AI-powered career trajectory modeling	Medium
🎨 Auto Diagram Generation	Convert text to visual flowcharts	Low

Phase 4: Institutional Integration (Q4 2026)

Feature	Description	Priority
🏫 LMS Integration	Connect with Moodle, Canvas, Blackboard	High
📚 Library API	Access university digital resources	Medium
👨‍🏫 Faculty Portal	Teacher dashboard for monitoring	Medium
📊 Institution Analytics	Department-wide performance insights	Medium
🎓 Graduation Pathway	Automated degree requirement tracking	Low

---

🐛 Challenges Overcome

Challenge	Solution
🔥 MCQ Quality	Fine-tuned T5 on domain-specific educational Q&A pairs + implemented diversity sampling and deduplication logic
⚡ RL Agent Convergence	Balanced reward function to prevent reward hacking + added entropy bonus for exploration
👥 Free-Rider Detection Accuracy	Combined multiple metrics (time, tasks, complexity) + statistical outlier detection with configurable thresholds
🔗 Module Integration	Microservices architecture with REST APIs + event-driven communication via Redis pub/sub
📄 PDF Text Extraction	Multi-library approach: pdf-parse for text, pdfjs-dist for scanned docs + text cleaning pipeline
🎯 Clustering Optimization	Elbow method for optimal k + silhouette score validation + feature engineering for better separation
⏱️ Real-time Performance	Redis caching layer + model quantization + asynchronous processing with worker queues
🔐 Data Privacy	End-to-end encryption + anonymized analytics + GDPR-compliant data handling

---

💰 Budget & Resources

Development Budget

Expense	Cost (LKR)	Justification
☁️ Cloud Hosting	8,000	Firebase (backend, database), Heroku (AI servers)
🌐 Domain & SSL	2,500	Professional web presence + security certificate
🤖 API Credits	5,000	LinkedIn/Indeed APIs, Google Calendar, OCR services
💾 Storage	3,000	PDF storage, model weights, user data
📚 Datasets	2,000	Educational Q&A corpus, job listings data
📄 Documentation	1,500	Reports, printing, presentation materials
Total	22,000	Reasonable for 4-person research project

Computational Resources

• Training: Google Colab Pro (free tier + paid sessions)
• Inference: Local servers + cloud deployment
• Storage: Firebase (5GB free) + AWS S3 (pay-as-you-go)

---

📚 References

Academic Papers

1. Karpicke, J. D., & Roediger, H. L. (2008). The critical importance of retrieval for learning. Science, 319(5865), 966–968.

2. Laban, C., et al. (2022). Question Generation: A Review. ACM Computing Surveys.

3. See, A., Liu, P. J., & Manning, C. D. (2017). Get to the point: Summarization with pointer-generator networks. ACL.

4. Pane, J. F., Steiner, E. D., Baird, M. D., & Hamilton, L. S. (2015). Continued Progress: Promising Evidence on Personalized Learning. RAND Corporation.

5. Keyes, C. L. M. (2007). Promoting and protecting mental health as flourishing. American Psychologist, 62(2), 95–108.

6. Schulman, J., Wolski, F., Dhariwal, P., Radford, A., & Klimov, O. (2017). Proximal Policy Optimization Algorithms. arXiv preprint arXiv:1707.06347.

7. MacQueen, J. (1967). Some methods for classification and analysis of multivariate observations. Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability, Volume 1: Statistics, 281--297.

Technical Documentation

8. Hugging Face. (2023). Transformers Library. https://huggingface.co/transformers

9. OpenAI. (2023). Reinforcement Learning with Human Feedback. https://openai.com/research/learning-from-human-preferences

10. Google. (2023). Google Calendar API Documentation. https://developers.google.com/calendar

---

🤝 Contributing

We welcome contributions from the community! Please read our Contributing Guidelines before submitting pull requests.

Development Workflow

1. Fork the repository
2. Create a feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

Code Standards

• Frontend: ESLint + Prettier
• Backend: Airbnb JavaScript Style Guide
• Python: PEP 8
• Testing: Jest (JS), pytest (Python)
• Documentation: JSDoc, docstrings

---

📄 License

This project is developed as part of academic research at SLIIT.

Project ID: 25-26J-228
Copyright © 2025 HelpMate Research Team

Licensed under the MIT License - see the LICENSE file for details.

---

🎓 Built with 💙 for Sri Lankan Students

🌟 Making Higher Education Smarter, Fairer, and More Supportive

---

⭐ Star this repo if you find it helpful!

📫 Contact Us

Email: helpmate.sliit@gmail.com
Website: helpmate.lk (Coming Soon)
LinkedIn: HelpMate Project

---