Fix SC/CP/DP postulate checkers and discussion-based overflow (#74 #75 #76 #80) - #98
Merged
Merged
Conversation
self_contradiction only checked that a self-attacking argument does not strictly outrank a clean one, so a tie passed. Bonzon et al. 2016 (p. 2, SC, checked against the page image) states (a, a) not in R and (b, b) in R imply a > b, strictly. The checker now requires every non-self- attacking argument to be strictly preferred to every self-attacker. Closes #76 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V4tAVyyKcs1sYcsEzL7Bzg
…tions cardinality_precedence skipped arguments with zero attackers and only compared arguments whose attackers were all unattacked, so ties on a plain a -> b -> c chain passed. Bonzon et al. 2016 (p. 2, CP, checked against the page image) states |R1-(a)| < |R1-(b)| implies a > b with no such restrictions. The checker now applies CP to every pair. Closes #75 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V4tAVyyKcs1sYcsEzL7Bzg
…r set defense_precedence treated an argument as defended only when every one of its attackers was attacked, so the paper's own partially defended example (h -> c -> a, d -> a, e -> b, g -> b) escaped the check. Amgoud & Ben-Naim 2013 (Example 3) and Bonzon et al. 2016 (p. 2, DP, checked against the page image) require |R1-(a)| = |R1-(b)|, R2+(a) nonempty and R2+(b) empty. The now-unused _every_attacker_is_attacked helper is removed. Closes #74 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V4tAVyyKcs1sYcsEzL7Bzg
discussion_based_ranking counted linear discussions with Python integers but converted every count to float, so the complete 3-cycle raised OverflowError at depth 650, and large counts lost precision before the lexicographic comparison. Amgoud & Ben-Naim 2013 (Defs. 10-11) define Dis_i(a) as exact counts compared lexicographically; the signed counts now stay integers. Closes #80 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V4tAVyyKcs1sYcsEzL7Bzg
The notes described Self-Contradiction as "ranked no higher" and limited Cardinality Precedence to non-attacked attackers. The page image (pngs/page-001.png, PDF p.2) states SC with a strict preference and CP as |R1-(a)| < |R1-(b)| => a > b with no such condition. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01V4tAVyyKcs1sYcsEzL7Bzg
# Conflicts: # papers/Bonzon_2016_ComparativeStudyRanking-basedSemantics/notes.md
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Fixes four ranking issues: three postulate checkers and discussion-based overflow. The postulate formulas were checked against the Bonzon et al. 2016 page image (PDF p. 2), not only the repo notes. Each commit adds a failing regression test (an all-tied ranking that the postulate forbids) plus a control (categoriser on the same framework), then applies the fix.
(a,a) ∉ R and (b,b) ∈ R ⇒ a ≻ b, a strict preference, but the checker allowed ties. It now requires every non-self-attacking argument to be strictly preferred to every self-attacker.|R1-(a)| < |R1-(b)| ⇒ a ≻ b. The checker's exemptions for zero-attacker arguments and for attacked attackers are removed._every_attacker_is_attackedhelper is removed.OverflowError.The branch also carries a docs commit correcting SC/CP in
papers/Bonzon_2016.../notes.md. Main had meanwhile corrected the same lines in b475c2f, and the merge takes main's version, so this PR adds no net notes change.Closes #74
Closes #75
Closes #76
Closes #80
🤖 Generated with Claude Code
https://claude.ai/code/session_01V4tAVyyKcs1sYcsEzL7Bzg