Qlink

AI powered mental model visualization with graph based link prediction using Qt6 and igraph.

Project Structure

├── code/           # Source code
│   ├── core/       # Core logic (models, AI, persistence, nlp)
│   ├── ui/         # User interface components
│   ├── tests/      # Test files
│   └── main.cpp    # Application entry point
├── docs/           # Documentation
└── README.md       # This file

Quick Start

# Install dependencies (macOS)
brew install qt6 igraph cmake

# Build and run
cd code
mkdir build && cd build
cmake ..
make -j8
./bin/Qlink                    # Run application
./bin/tests/qlink_tests        # Run all tests

See docs/INSTALL.md for the installation instructions

Major Features Implemented in this Deliverable

1. Command Pattern with Full Undo/Redo Support

Location: core/nlp/Commands.h and core/nlp/Commands.cpp

• AddConceptCommand: Stores added concept ID for undo
• RemoveConceptCommand: Stores removed concept AND all its relationships for complete restoration
• CreateRelationshipCommand: Stores relationship ID for undo
• DeleteRelationshipCommand: Stores entire relationship object for restoration

All commands now fully implement:

• execute(): Performs the operation and stores state
• undo(): Reverses the operation using stored state
• getDescription(): Returns human-readable description

Tests: 24 comprehensive tests in tests/nlp/test_Commands.cpp (all passing)

2. Natural Language Command Factory

Location: core/nlp/CommandFactory.h and core/nlp/CommandFactory.cpp

Implemented sophisticated regex-based NLP parser with 5 patterns:

1. Add/Create Concept: "add concept AI with description 'Artificial Intelligence'"
2. Remove/Delete Concept: "remove concept AI"
3. Connect Concepts: "connect AI to ML with type 'related' directed"
4. Disconnect Concepts: "disconnect AI from ML"
5. Simple Add: "add AI" (shorthand)

Features:

• Case-insensitive matching
• Support for quoted names and descriptions
• Optional parameters (description, type, directed flag)
• Concept name resolution (finds by name, not just ID)
• Helper functions: trim(), toLower(), findConceptIdByName()

Tests: 28 comprehensive tests in tests/nlp/test_CommandFactory.cpp (all passing)

3. Modern UI Design

Location: ui/MainWindow.cpp - applyModernStyling()

Applied comprehensive Qt stylesheet with:

• Dark menu bar (#2c3e50) with hover effects
• Modern toolbar (#34495e) with rounded buttons
• Professional color scheme: Blues (#3498db), grays (#ecf0f1, #bdc3c7)
• Rounded corners (border-radius: 4-6px) on all interactive elements
• Hover states for better user feedback
• Smooth transitions and visual depth
• Dock widget styling with custom title bars

4. Natural Language Command Panel

Location: ui/MainWindow.cpp - setupNaturalLanguagePanel()

New QDockWidget added to bottom of main window with:

• Command input: Multi-line QTextEdit for natural language commands
• Help text: Examples of supported commands
• Execute button: Styled with primary blue color (#3498db)
• Clear History button: For cleaning command log
• Command history: QListWidget showing:
  • ✓/✗ status indicators with color coding (green/red)
  • Timestamp for each command
  • Original command text
  • Success/error messages
  • Auto-scrolling (newest first)
  • Limited to 50 entries

Features:

• Real-time validation using CommandFactory::isValidCommand()
• Detailed error messages with command examples
• Automatic input clearing on success
• Integration with model change signals
• Professional styling matching overall UI theme

6. Test Infrastructure Improvements

Location: tests/CMakeLists.txt

Simplified test build configuration:

• Removed gtest_discover_tests() (no CTest logs)
• Removed gmock dependencies (not needed)
• Changed to GLOB_RECURSE for automatic test discovery
• Updated library linking: Assignment2Core → QlinkCore

Test Results

All 158 tests passing across 8 test suites:

• 22 Concept tests
• 27 Relationship tests
• 27 MentalModel tests
• 11 CommonNeighborPredictor tests
• 9 JaccardCoefficientPredictor tests
• 10 PreferentialAttachmentPredictor tests
• 28 CommandFactory tests
• 24 Commands tests

Design Patterns Used

1. Command Pattern: Execute/undo with state preservation
2. Factory Pattern: CommandFactory for creating commands from strings
3. Strategy Pattern: Link prediction algorithms
4. Observer Pattern: Qt signals/slots for model changes
5. Singleton-like: Model management through unique_ptr

User Experience Improvements

Before

• Basic UI with default Qt styling
• No natural language input
• Manual menu navigation for all operations
• Limited visual feedback

After

• Modern, professional appearance with rounded corners and shadows
• Natural language command panel with history
• Real-time command validation and feedback
• Color-coded success/error indicators
• Comprehensive help text and examples
• Auto-scrolling command history
• Dark menu bar and toolbar for better contrast

Command Examples

Users can now type natural language commands:

add concept AI with description 'Artificial Intelligence'
add concept ML
connect AI to ML with type 'related'
connect ML to DL directed
remove concept AI
disconnect ML from DL

Documentation

• Installation Guide - Setup instructions for different platforms

Development

Built with:

• Qt6 for cross-platform GUI
• igraph for graph algorithms
• C++17 with modern practices
• CMake for build system