Apply ZIP update: 205-didactopus-learner-state-progression-update.zip [2026-03-14T13:20:33]

pyproject.toml

@@ -6,18 +6,7 @@ build-backend = "setuptools.build_meta"
 name = "didactopus"
 version = "0.1.0"
 requires-python = ">=3.10"
-dependencies = [
-  "pydantic>=2.7",
-  "fastapi>=0.115",
-  "uvicorn>=0.30",
-  "sqlalchemy>=2.0",
-  "passlib[bcrypt]>=1.7",
-  "python-jose[cryptography]>=3.3"
-]
-
-[project.scripts]
-didactopus-api = "didactopus.api:main"
-didactopus-export-svg = "didactopus.export_svg:main"
+dependencies = ["pydantic>=2.7", "pyyaml>=6.0"]
 
 [tool.setuptools.packages.find]
 where = ["src"]

src/didactopus/progression_engine.py

@@ -1,6 +1,10 @@
 from __future__ import annotations
+from datetime import datetime, timezone
 from .learner_state import LearnerState, EvidenceEvent, MasteryRecord
 
+def _parse_ts(ts: str) -> datetime:
+    return datetime.fromisoformat(ts.replace("Z", "+00:00"))
+
 def apply_evidence(
     state: LearnerState,
     event: EvidenceEvent,
@@ -19,13 +23,34 @@ def apply_evidence(
         )
         state.records.append(rec)
 
+    prev_score = rec.score
+    prev_conf = rec.confidence
+
+    # weighted incremental update
     weight = max(0.05, min(1.0, event.confidence_hint))
-    rec.score = ((rec.score * rec.evidence_count) + (event.score * weight)) / max(1, rec.evidence_count + 1)
+    rec.score = ((prev_score * rec.evidence_count) + (event.score * weight)) / max(1, rec.evidence_count + 1)
+
+    # confidence grows with repeated evidence and quality, but is bounded
     rec.confidence = min(
         1.0,
-        max(0.0, rec.confidence * (1.0 - decay) + reinforcement * weight + 0.10 * max(0.0, min(1.0, event.score))),
+        max(
+            0.0,
+            prev_conf * (1.0 - decay) + reinforcement * weight + 0.10 * max(0.0, min(1.0, event.score))
+        ),
     )
+
     rec.evidence_count += 1
     rec.last_updated = event.timestamp
     state.history.append(event)
     return state
+
+def decay_confidence(state: LearnerState, now_ts: str, daily_decay: float = 0.0025) -> LearnerState:
+    now = _parse_ts(now_ts)
+    for rec in state.records:
+        if not rec.last_updated:
+            continue
+        then = _parse_ts(rec.last_updated)
+        delta_days = max(0.0, (now - then).total_seconds() / 86400.0)
+        factor = max(0.0, 1.0 - daily_decay * delta_days)
+        rec.confidence = max(0.0, rec.confidence * factor)
+    return state

src/didactopus/recommendations.py

@@ -13,7 +13,14 @@ def recommend_next_concepts(
     for concept in concepts:
         cid = concept.get("id")
         prereqs = list(concept.get("prerequisites", []) or [])
-        ready = concept_ready(state, cid, prereqs, dimension=dimension, min_score=min_score, min_confidence=min_confidence)
+        ready = concept_ready(
+            state,
+            cid,
+            prereqs,
+            dimension=dimension,
+            min_score=min_score,
+            min_confidence=min_confidence,
+        )
         if ready:
             existing = state.get_record(cid, dimension)
             if existing is None or existing.score < min_score or existing.confidence < min_confidence: