11/6/2026 - Course Notebooks Read the Game: Argumentum in CoursIANews | Argumentum
Skip to main content
Home  › ... News

Course Notebooks Read the Game: Argumentum in CoursIA

11/6/2026 Evolution
Course Notebooks Read the Game: Argumentum in CoursIA

CoursIA, a public repository of course materials on artificial intelligence, devotes a series of notebooks to argumentative analysis. Five of them take the Argumentum corpus as their terrain — taxonomy, ontology, cross-links, virtues — on a version of the game’s repository frozen expressly for them.

Illustration: docs/publication/articles/illustrations/4a-cinq-carnets.png, Argumentum repository, commit 981e5233 — typographic montage based on the CoursIA repository at commit 7eea21ebefe: the five notebooks that consume the corpus, in their actual folder, with the submodule pin.

The repository and its series

A Jupyter notebook is a document that mixes text and code that runs step by step: each step is displayed, edited, and replayed. The series is called Argument_Analysis — “argumentative analysis” — and includes 38 notebooks, from the Toulmin model, which breaks an argument down into its elements, to Dung's semantics, the mathematics that tells which arguments in a debate are left standing.

Its method is documented: a language model extracts the argumentative fabric of a text — premises, conclusions, transitions; a SAT solver, a program that decides whether logical formulas can be satisfied together, checks the consistency of the formalization; an orchestration layer assembles everything into a reproducible pipeline.

A frozen version, for replay

Argumentum comes in as a submodule: a reference recorded in the CoursIA repository that points to the game's repository at a specific commit. The pin points to the November 5, 2026 commit. This is what guarantees that a notebook written today will run tomorrow on the same corpus: the taxonomies don't shift under the teacher's feet, and every dated measurement remains replayable.

The notebooks that read the game

Five notebooks consume the corpus directly:

  • Detection — the taxonomy of fallacies serves as the detector's reference: each known label is a node in the tree.
  • The cards — the game's deck is derived from the taxonomy: from the tree node to the playable card, by way of the illustrations.
  • The ontology of fallacies — it reads argumentum_fallacies.owl, the formal export of the taxonomy. The standard reading libraries, rdflib and owlready2, fail on this format: the notebook therefore builds its own tolerant parser, and documents this fragility as that of the bridge, not of the file.
  • Cross-links — this one reads the canonical CSV, with its eight columns of relations between fallacies and its alignments to Walton's argumentation schemes, a classic typology of presumptive reasoning. It counts 116 alignments, and measures the agreement between the CSV and the ontology at the pin: across 145 attack relations and 116 alignments, it is total.
  • The virtues — this one opens argumentum_virtues.owl, the game's positive pole, as a SKOS thesaurus, a standard vocabulary for publishing classifications: 2,897 triples, root "Valid argument".

What reading reveals

These notebooks do not merely consume: they measure, and their measurements teach the corpus itself things. First lesson, quantified by the cross-links notebook: the game's curation effort goes into the tree — eight languages, seven families, ten levels — far more than into the links between its branches, whose density is measured at 4.79%. The graph above the tree remains largely to be written. Second lesson, recorded as a gap to be reconciled: two ontology files coexist in the readings — one local to the series, the other from the pinned submodule — and they do not carry the same layer of cross-links. This is the flip side of a living corpus: each pin has its own states, and the notebooks date them.

Traffic in both directions

The path doesn't only lead from the game to the courses. Three tools born in the game's repository — an ontology adapter, a CSV file comparison engine for synchronizing with a spreadsheet, and the prompt set that drives the translation of the corpora — were brought up into CoursIA as exercise materials. Tracking of the whole is public and open; the series declares itself at "beta" maturity for 36 of its 38 notebooks. A living course, in short — one that learns as much from the corpus as it teaches it.


Sources: CoursIA repository, commit 7eea21eb (November 8, 2026): series description, submodule declaration, and the five notebooks linked above — the counts they establish (116 alignments, 145/145, 2,897 triples, 4.79%, 38 notebooks of which 36 in "beta") are read at this commit. Argumentum submodule pinned at commit 4332d26a (November 5, 2026). Illustration: Argumentum repository, commit 32408fdd. Tracking: Epic CoursIA #4960.

‹ Back to the list

Contact Us

Newsletter Sign-Up

We were unable to confirm your subscription.
Your subscription has been confirmed.

Newsletter

Subscribe to our newsletter to keep up with the latest news from Argumentum.

We use Sendinblue as our marketing platform. By submitting this form, you acknowledge that the information you provide will be transferred to Sendinblue in its capacity as a data processor, in accordance with its terms and conditions of use.