✈️ Airport Finder (Semantic)

Problem Statement

Searching for airports is usually limited to exact string matches (e.g., typing “Heathrow” or “JFK”). This doesn’t work well when users only have vague descriptions (“major airport near New York”), natural language queries (“tropical hub in Southeast Asia”), or incomplete information.

Travelers, airlines, and logistics companies need a smarter way to discover airports and routes without relying on exact codes or names.

---

Proposed Solution

Airport Finder (Semantic) enables natural language search over the OpenFlights airport dataset by combining MariaDB Vector features with Python.

• Use sentence embeddings to represent airport metadata (name, city, country).
• Store embeddings in MariaDB VECTOR columns with ANN indexing.
• Query airports by semantic similarity instead of string match.
• Provide a simple API + UI to allow searching for airports in plain English.

Example query:

“Find me an international hub in South India close to a coastal city.”
Result: Returns airports like Chennai (MAA) or Kochi (COK).

---

Why MariaDB?

• VECTOR datatype and ANN indexes enable fast semantic search.
• Native distance functions (cosine, euclidean) for ranking results.
• Seamless integration with Python, FastAPI, and Jupyter notebooks.
• Open-source and scalable, making it perfect for real-world travel data.

---

Architecture Overview

User Query → FastAPI (Python) → Embedding Model (MiniLM-L6-v2)
           → MariaDB (VECTOR search on OpenFlights data)

Expected Impact

• Travelers: Discover suitable airports even without knowing exact codes/names.
• Airlines / Logistics: Identify alternative hubs quickly.
• Developers: Example project showcasing MariaDB’s vector search for semantic retrieval.

---

Feasibility & Next Steps

Idea Phase (Sep 2025)

• Load OpenFlights data into MariaDB.
• Add vector embeddings for airport text fields.
• Validate semantic queries with sample prompts.

Prototype Phase (Oct 2025)

• Build FastAPI search service.
• Create a map-based UI to visualize airport search results.
• Record demo video (2–4 minutes).
• Publish public GitHub repo with open-source license.

---

Team

• Solo project (Namish M.) for the MariaDB Python Hackathon 2025.

---

✨ In One Line

Airport Finder (Semantic) lets you search the world’s airports in natural language, powered by MariaDB Vector.

       → Top-K Similar Airports → API/UI (Map Visualization)