UDO is a human-centric recommendation engine that prioritizes user agency and data ownership. Instead of black-box predictions, UDO empowers users to:
- Write and update primary preferences, secondary desires, and personal notes
- Apply likes, scores, or flags to items
- See and revise the full list of their inputs
- Explore transparent, adaptive recommendations based on their own logic
Motto: User Data Ownership
-
Expressive input system
Users can define personal tastes and multi-layered desires using structured free-text. -
Reversible feedback
Likes, scores, or reactions are not final — users can review and reset them. -
Transparent logic
A "Personalized Recommendations" tab lets users trace why a suggestion appears. -
You own your data
UDO stores data locally or in a user-controlled backend. Nothing is hidden or harvested. -
Composable architecture
Designed for integration with marketplaces, media apps, or any user-facing system.
UDO is built around 4 core components:
-
Input Engine
Structured fields and open text (primary/secondary/... preferences, contextual info). -
Feedback Tracker
Likes, dislikes, scores, emotional tags — all accesible, editable and reviewable. -
Recommendation Core
Uses LLM + rules-based filtering to propose suggestions from a public or private catalog. -
User Dashboard
View, revise, or remove inputs; see how recommendations change (in real time?).
At this stage, the project is in early development. We welcome discussions, ideas, and issue reporting.
If you’d like to collaborate, please open an issue.
UDO isn't just a recommender — it's a mirror of your intent.