fix(tweety): reformat single-line cells + Papermill re-exec Tweety-5 (#1946)

[jsboige]

Summary

Convention #1946 : fix source formatting + re-execution Tweety-5-Abstract-Argumentation.

Problem

10 code cells (5 exercise stubs + 5 solved examples) had their source stored as single-line compressed strings with no newlines in the JSON, making them syntactically broken. All 10 cells also had set but empty .

Changes

1. Source formatting fix : Reformatted 10 cells from single-line to proper multi-line source

   • 5 exercise stubs (ex1-ex5) : preserved
   • 5 solved examples (sol-ex1 to sol-ex5) : proper Python with newlines

2. Full Papermill re-execution : All 19 code cells now have outputs, 0 silent cells

3. Installed jpype in conda env coursia-ml-training (was missing, required for Tweety JVM)

Validation

• Papermill SUCCESS (59/59 cells, 9s execution time)
• 0 silent code cells (convention docs(rules): code cells emit informative output (anti false-positive maturity) #1946)
• No local paths in outputs
• generate_catalog.py --git-tracked-only : 504 entries
• Source changes : 10 cells (insertions 481 > deletions 311)

Files changed

File	Change
Tweety-5-Abstract-Argumentation.ipynb	10 cells reformatted + all outputs re-generated
COURSE_CATALOG.generated.json	Regenerated (504 entries)
SymbolicAI/README.md	Catalog marker updated

[github-actions]

Notebook PR Validation: PASS

• Notebooks checked: 1
• Code cells validated: 19
• Result: All passed

Checks: H.1 (no errors), H.3 (execution_count), C.1 (no banned patterns)
Non-Python kernels (.NET/Lean): C.1 + errors only (execution_count advisory)
QuantConnect notebooks: C.1 + errors only (require QC Cloud for execution)

[clusterManager-Myia]

[Hermes] — COMMENT_WITH_CONCERNS

Security scan: CLEAN. Notebook validation: PASS.

Concerns:

1. CI FAIL — Notebook catalog drift. Le catalog drift check echoue. Le catalog JSON/MDFinal inclut des promotions BETA→PRODUCTION et des changements de metadata (status DEMO→READY, cells_with_outputs 9→14 pour SC-0) qui ne sont pas en phase avec le catalog attendu.

2. Scope: catalog + Tweety-5 re-exec + SymbolicAI README. Trois types de changements dans une PR. Le Tweety-5 re-exec est papermill standard (timestamps mis a jour, random seed different donc outputs differents). La README SymbolicAI met a jour les compteurs de maturation. RAS sur le contenu mais le catalog drift doit etre resolu.

3. Tweety-5: nouveau print fragment. Le stdout du cell de verification JVM est splitte en deux output streams ("JDK portable: zulu17..." dans un bloc, "JVM demarree avec 42 JARs." dans un autre). Probablement un effet de papermill, mais verifier que ca n'''impacte pas la lecture du notebook.

[jsboige]

[NanoClaw] COMMENT

Security scan: CLEAN. No leaked secrets, tokens, IPs, or paths. Papermill metadata paths fixed from Windows absolute (D:\dev\CoursIA...) to relative paths — correct.

Notebook validation (Tweety-5-Abstract-Argumentation.ipynb):

Source formatting fix — 10 cells reformatted from single-line to proper multi-line. Source content is identical, only whitespace/newlines changed. Correct.

Outputs (5 solved examples):

• sol-ex1 (diamond AF): Grounded={a,c}, Preferred=1, Stable=1 — consistent
• sol-ex2 (5-cycle): Stable=0, CF2=5 extensions — correct (odd cycle has no stable)
• sol-ex3 (random stats, N=100): 92/100 with stable, avg grounded 0.76, avg preferred 1.65 — plausible, not round numbers
• sol-ex4 (debate AF): Grounded shifts {A,C,D} -> {A,C,E} after adding E — correct
• sol-ex5 (fully symmetric 4-arg): Grounded={} — correct for complete symmetry

Outputs (5 exercise stubs): All print placeholder text, no fake outputs.

Papermill metadata: All cells re-executed 2026-05-31T22:54:55-22:55:02 UTC (~9s window). Python 3.12.13 (was 3.13.12). Duration 9.28s (was 3.47s, consistent with jpype/JVM startup). Paths fixed. No concerns.

Catalog (COURSE_CATALOG.generated.json): 25+ entries updated with new last_validation dates and issue references. Reflects multiple merged PRs (#1995, #1985, #1990, #1997, #2001, #2008, #1984). Includes maturity promotions (DEMO->READY, BROKEN->READY) and cells_with_outputs update for Tweety-5 (9->14). This is broader than Tweety-5 alone — catalog regen after rebase on main.

No concerns on the notebook changes. The catalog drift CI concern from the previous review depends on whether CI is now aligned with main — cannot verify without CI access.