Z-ENG: Ingredient-Aware Skincare Routine Recommender
2025-2026 tavasz
Szoftver
Téma leírása
Part 1: Project Lab (Design & MVP)
Design and implement an ingredient-aware skincare routine recommendation system that helps users build a simple AM/PM routine based on their skin type, sensitivity, and preferences, and then suggests suitable products from a real dataset.
The system should behave like a practical “decision-support” tool (not a medical diagnosis app). It should guide the user to a minimal routine and explain recommendations in a clear way.
Main requirements include:
- User profile management (skin type, goals, sensitivity flags, budget, preferred routine simplicity)
- Product database ingestion from a public cosmetic dataset (products + ingredient lists)
- Ingredient-aware filtering (e.g., avoid user-selected irritants such as fragrance)
- Routine generator that outputs a basic AM/PM sequence (cleanser → treatment (optional) → moisturizer → sunscreen in the morning)
- Recommendation engine that ranks top products for each routine step using at least:
- a rule-based baseline (filters + scoring), and
- a content-based method (ingredient similarity)
- Explainable outputs
- Local database for storing products, user profiles, and saved routines/favorites
- Simple graphical interface (web or desktop) where users can enter preferences, see routines, and save favorites
The outcome of this part is a working minimum viable product (MVP) that can take a user profile and produce a reasonable routine with ranked product suggestions and basic explanations, tested on a small sample of products.
Part 2: Thesis (Enhancements & Evaluation)
Extend the system to be more robust, more personalized, and more academically evaluated.
Possible enhancements include:
- Hybrid recommendation (combining rules + similarity + popularity/coverage balancing)
- Improved user feedback loop (likes/dislikes influencing future ranking)
- Reporting and export (PDF/HTML routine summary, product lists, shopping list)
- Model evaluation using standard recommendation metrics
- Usability improvements (validation, clear warnings, better UI design)
Final deliverables include a functional and tested application, source code, documentation, and a short evaluation discussing recommendation quality, design choices, limitations, and future improvements.
Maximális létszám:
1 fő