This project was developed as part of the Master's in Informatics Engineering at ISCTE-IUL. It addresses the issue of land fragmentation in Portugal by providing a web-based tool that uses mathematical algorithms to propose fair and equitable land exchanges.

The goal is to help regional decision-makers and owners optimize territory distribution, reducing environmental degradation and rural fire risks associated with dispersed, abandoned properties.

• Alexandre Milharado (105946)
• Marco Camargo (128933)
• Pedro Moura (126474)
• Sara Santos (131804)

The application operates as a Minimum Viable Product (MVP) with four main modules:

1. Authentication: User session management (login/logout) via cookies.
2. Simulation:
   • Upload of geographic data (.gpkg format).
   • Simulation of land ownership using Uniform or Poisson distributions.
3. Defragmentation: Execution of optimization algorithms to reorganize land parcels.
4. Visualization: Interactive map views of the initial state versus the optimized result, including statistical reports on exchanges.

The solution utilizes a decoupled architecture

• Backend: Django (Python)
  • Handles data processing and algorithm execution.
  • Uses GeoPandas for processing geographic datasets
• Frontend: Angular (TypeScript)
  • Single-Page Application (SPA) structure.
• Styling: Tailwind CSS.

We developed three distinct algorithms to solve the optimization problem, balancing aggregation error against area preservation.

A heuristic approach that effectively "wipes" the map and redistributes land in consolidated blocks based on the owner's total initial area .

• Pros: Extremely fast execution; maximizes aggregation. Cons: Does not account for original plot locations; requires logic to handle leftover unallocated land.

An evolutionary algorithm combining Beam Search and Simulated Annealing Mechanism: Uses genetic operators (swap, multiple swap, cluster expansion) to mutate states. It maintains a "beam" of best solutions but allows accepting worse states (based on temperature) to escape local minima. Performance: Achieved the lowest aggregation error in testing (best quality results)

Inspired by Tabu Search and Informed Greedy Search. Mechanism: Identifies "pivot" plots (largest/best aggregated) for each owner and attempts to expand around them by trading or "buying" neighboring plots from other owners. Pros: Preserves the logic of the original location better than redistribution

Tests were conducted using IFAP datasets (e.g., Portalegre, Lisboa, Vila Real).

Data based on project report results.

• Value Definition: Incorporate economic factors (soil quality, road access, taxes) rather than just area size for fairer swaps.
• Architecture: Migrate from the current monolithic backend to a microservices architecture using RabbitMQ for asynchronous processing of heavy algorithms.

This project uses data from IFAP (Instituto de Financiamento da Agricultura e Pescas)