Papers by Valérie Aldana
Discrepancy Detection at the Data Level: Toward Consistent Multilingual Question Answering (2025.emnlp-main)
Copied to clipboard
Lorena Calvo-Bartolomé, Valérie Aldana, Karla Cantarero, Alonso Madroñal de Mesa, Jerónimo Arenas-García, Jordan Lee Boyd-Graber
| Challenge: | Multilingual question answering systems must ensure factual consistency across languages while also accounting for cultural variation in subjective responses. |
| Approach: | They propose a user-in-the-loop fact-checking pipeline to detect factual and cultural discrepancies in multilingual QA knowledge bases. |
| Outcome: | The proposed tool detects factual and cultural discrepancies in bilingual question answering systems. |