Multilingual extension of the CAP framework

Develop a multilingual extension of the CAP framework for semantic correctness evaluation by using multilingual natural-language-inference checkpoints to port the framework beyond English.

Background

The experiments evaluate CAP on English educational question-answering datasets, and the limitations section notes that the benchmark and current framework are restricted to English. The authors explicitly identify multilingual NLI checkpoints as an open direction for extending CAP’s applicability to languages other than English.

References

Several directions remain open for future work. Splitting answers into smaller subcomponents and scoring entailment over each could resolve the bidirectional-NLI artefact on the overinclusive-valid / partial axis; stronger NLI backbones with broader world knowledge could close the alternative-correct ceiling; and multilingual NLI checkpoints could port the framework off English.

— How Correct Is Your Answer? A Semantic Correctness Framework for Open QA Evaluation  (2609.01369 - Yotkova et al., 1 Sep 2026) in Section 7, Conclusion and Future Work