Skip to content

User Profile Schema (Canonical)

Canonical location for User Profile specification. Supersedes ../specs/user_profile_schema.md.


1. Core Identity

Field Type Description
user_id string (GUID) Unique identifier
name string Display name
email string Login / notifications
language string Preferred UI language (BCP 47)
location string Home base / default region
created_at datetime Account creation
last_active datetime Engagement / churn signal

2. Saved Content

Field Type Description
saved_petals string[] Bookmarked petals
saved_trips string[] Stored trip IDs
favorite_tags string[] Preference weighting
recent_searches string[] Last N queries (rotating)
discovered_petals string[] Found spontaneously
missed_petals string[] Planned but skipped

3. Preferences

Field Type Description
travel_modes string[] campervan, bike, car, walk
companions string[] kids, partner, dog, solo
seasonality_pref string[] Preferred seasons
accessibility_needs string[] e.g. wheelchair-friendly
notification_opt_in boolean Messaging enable

4. Feedback & Ratings

Field Type Description
petal_feedback object[] { petal_id, rating(1..5), comment }
trip_feedback object[] { trip_id, rating(1..5), comment }
suggested_petals object[] User idea seeds

5. Personalization Signals

Signal Source Usage
favorite_tags explicit saves Boost matching petals
discovered_petals trip logs Exploration expansion
missed_petals trip logs Make-up suggestions
petal_feedback.rating feedback form Ranking label
seasonality_pref profile Seasonal biasing

6. Embeddings & Privacy

  • Optional preference_embedding
  • Truncate / hash old recent searches (privacy)
  • PII minimization for future regulatory scope

7. Caching Keys

Purpose Pattern TTL
Profile core profile:{user_id}:v1 5m
Preference vector profile:vec:{user_id}:v1 30m

8. RAG Use Cases

  • Inject top preference tags + companions into context summary
  • Use missed/discovered petals for diversification
  • Use feedback sentiment to prune low-quality petals

9. Future Considerations

  • Federated personalization
  • Time-decay weighting
  • Multi-tenant partitioning

See also: petal_schema.md, trip_schema.md, ../architecture_new/data_flow_rag.md

Last updated: September 2025