Enthymeme Detection & Proposition Generation — MediaEval 2026
Centre for Language Studies, Radboud University Nijmegen
The following schemes are the argumentation patterns most frequently observed in the dataset. Each entry shows the abstract logical form followed by an annotated example drawn from the data. In each example, the implicit element — the proposition requiring annotation — is highlighted in orange. Explicit premises appear in green and conclusions in blue.
In the abstract forms, single-letter tokens such as A, G, P, X stand for schematic variables and are printed in italics.
| P1 | Action A violates moral principle P |
| P2 | Agent X performed action A |
| Conclusion | Therefore, Agent X acted wrongly / should be condemned |
| P1 | Agent A now claims / supports principle P |
| P2 | Agent A previously violated or denied principle P |
| P3 | Previously violating P while now claiming P is hypocritical |
| Conclusion | Therefore, Agent A is hypocritical / inconsistent |
| P1 | Actor A claims that X causes or signals Y |
| P2 | X does not cause or signal Y; the connection is false |
| Conclusion | Therefore, Actor A is misleading or manipulating the audience |
| P1 | Party P supports policies X |
| P2 | Policies X are harmful to group G |
| Conclusion | Therefore, Party P has abandoned / harmed group G |
| P1 | Category C exhibits characteristics X |
| P2 | Agent A exhibits characteristics X |
| Conclusion | Therefore, Agent A belongs to / is acting like category C |
| P1 | Goal G is good / necessary |
| P2 | Action A promotes or achieves G |
| Conclusion | Therefore, action A is good / necessary |
| P1 | Standard S requires condition C for X to be acceptable |
| P2 | X does not meet condition C |
| Conclusion | Therefore, X is unacceptable / should not be trusted |
| P1 | We have Y (substitute) |
| P2 | Y produces the same effect as X |
| Conclusion | Therefore, you do not need X |
| P1 | If Agent A can perform harmful action X, then A cannot be trusted / must be stopped |
| P2 | Agent A can perform harmful action X |
| Conclusion | Therefore, Agent A cannot be trusted / must be stopped |