Project Overview

This project, conducted in the context of the Winter Semester 2025/26 “Hands-on Recommender Systems” course at the Technical University of Munich (TUM), aims to enhance outdoor activities such as hiking, cycling, and skiing by developing an intelligent, context-aware, and gamified route recommendation platform. The project was carried out collaboratively by Kexin Lu, Tianhao Gu, and Jin Shi.

Motivation & Objectives

Travelers increasingly suffer from information overload when choosing outdoor adventures. Our goal was to design a personalized tourism recommender system that intelligently adapts suggestions based on user context, preferences, and environmental conditions, while boosting motivation and engagement through gamification.

Methodology

• Hybrid Recommendation Engine: Combines personalized, context-aware filtering (current weather, user location, experience level, preferred duration, etc.) with real-time feedback and gamified progression.
• Gamification: Users complete sequences of route challenges. Finishing a route unlocks subsequent, more difficult challenges, creating a ladder of outdoor adventures. Completing tasks grants experience points (XP), with narrative-driven elements and mini-quests enriching the journey.
• Feedback Loop: User feedback (likes, dislikes, difficulty reports) is continually integrated to refine recommendation quality and personalization.
• Data Sources: Primarily uses the Outdooractive API, supplemented by open datasets for context enrichment.
• Evaluation: System performance is measured by established recommender metrics, including Precision@k, Recall, and MAP.

Deliverables

• Web-based Interactive Prototype: Demonstrates real-time, hybrid route recommendation and adaptive gamification flow.
• Milestone & Final Presentations: Regular milestone reports and a final presentation to peers and faculty.
• Comprehensive Final Report: Documented functionality and experimental findings in ACM-compliant format (single column, max 10 pages), with detailed GitHub contributions and annotated screenshots of the UI.

My Role

• Co-developed the recommendation algorithms and feedback analysis pipeline
• Designed and implemented core web UI and adaptive gamification logic
• Coordinated dataset pipelines and presentation preparation

Key Takeaways

• Hands-on experience building a scalable recommender platform for real-world, user-driven contexts
• Deeper understanding of hybrid AI approaches, user experience design, and collaborative development in cross-university teams

Contact: Kexin Lu, Tianhao Gu, Jin Shi, TUM, Germany