Get Started
Home
Topics
Search
Library
Research questionHow can systems fact-check multilingual spoken dialogue when claims span turns and ASR distorts the text?A factual claim in conversation may depend on earlier turns or multiple speakers, while transcription errors can change the text that must be checked. Verification also requires finding relevant source evidence rather than judging an isolated sentence.
Audio & Speech
Evaluation & Benchmarks
Information Retrieval
Natural Language Processing
Latest papersRecent research connected to this question, newest first.TRILOGUE: A Trilingual Spoken Dialogue Fact-Checking Benchmark with Evidence and Paired AudioThe evidence covers English, Russian, and Kazakh dialogues with paired audio, ASR transcripts, word-level timestamps, turn-level labels, and source articles. It supports claim check-worthiness detection, evidence retrieval, and verification using claim-only, gold-evidence, or retrieved-evidence inputs; reported baselines show ASR degradation and cross-lingual transfer remain difficult, particularly for Kazakh.research paper · Sep 3, 2026
Related questions
How can multilingual NLP models reliably detect harmful posts and verifiable factual claims in low-resource languages?How can long-term conversational QA agents retrieve and reason over temporally dispersed dialogue history?How can high-stakes LLM systems distinguish unsupported claims from novel ones and prioritize expert verification?How reliably do vision-language models correct repeated visually grounded false premises across dialogue turns?