#16 - Implement: SM-2 review engine — scheduling, due cards, and rating - gaetan/knowledgify

**Wayfinder map:** #1 **Dependencies:** #13 (project scaffolding), #15 (database layer) ## Description Implement the SM-2 spaced repetition algorithm: query due cards, apply ratings, update scheduling. This is the core learning engine. ## Tasks - [ ] Create `internal/review/` package - [ ] Implement SM-2 algorithm: - `new_interval = old_interval * ease_factor * adjustment` - Adjustment factors: Again=0, Hard=1.2, Good=2.5, Easy=4.0 - Ease factor adjustment: `ease_factor = ease_factor + (0.1 - (5 - rating) * (0.08 + (5 - rating) * 0.02))` - If rating is "Again", reset interval to 1 day - [ ] Rating enum/types: - `Again` (0) — card not remembered, reset to 1 day - `Hard` (1.2) — partial recall - `Good` (2.5) — full recall - `Easy` (4.0) — easy, increase interval more - [ ] Query due cards: `WHERE next_review <= now` - [ ] Prioritization: new cards first, then by next_review ascending - [ ] Max 10 new cards per day (configurable) - [ ] Interval cap at 30 days - [ ] Update progress after each rating - [ ] Session management: get next card, rate card, check session complete - [ ] Unit tests for SM-2 algorithm with known inputs/outputs ## Acceptance criteria - SM-2 formula produces correct intervals for known inputs - Due cards are returned in correct priority order - Max 10 new cards per day is enforced - Interval capped at 30 days - Rating updates progress correctly - Unit tests pass with verified SM-2 calculations

gaetan commented

2026-08-20 15:31:53 +02:00

Owner

Wayfinder map: #1
Dependencies: #13 (project scaffolding), #15 (database layer)

Description

Implement the SM-2 spaced repetition algorithm: query due cards, apply ratings, update scheduling. This is the core learning engine.

Tasks

Create internal/review/ package
Implement SM-2 algorithm:
- new_interval = old_interval * ease_factor * adjustment
- Adjustment factors: Again=0, Hard=1.2, Good=2.5, Easy=4.0
- Ease factor adjustment: ease_factor = ease_factor + (0.1 - (5 - rating) * (0.08 + (5 - rating) * 0.02))
- If rating is "Again", reset interval to 1 day
Rating enum/types:
- Again (0) — card not remembered, reset to 1 day
- Hard (1.2) — partial recall
- Good (2.5) — full recall
- Easy (4.0) — easy, increase interval more
Query due cards: WHERE next_review <= now
Prioritization: new cards first, then by next_review ascending
Max 10 new cards per day (configurable)
Interval cap at 30 days
Update progress after each rating
Session management: get next card, rate card, check session complete
Unit tests for SM-2 algorithm with known inputs/outputs

Acceptance criteria

SM-2 formula produces correct intervals for known inputs
Due cards are returned in correct priority order
Max 10 new cards per day is enforced
Interval capped at 30 days
Rating updates progress correctly
Unit tests pass with verified SM-2 calculations

gaetan referenced this issue

2026-08-20 15:32:07 +02:00

Implement: HTTP handlers and routing — dashboard, categories, and review endpoints #17

Rows
Columns

Implement: SM-2 review engine — scheduling, due cards, and rating #16

Description

Tasks

Acceptance criteria