SemQA: Evaluating Evidence with Question Embeddings and Answer Entailment for Fact Verification

Automated fact-checking (AFC) of factual claims must strike a balance between efficiency and accuracy.Although sophisticated frameworks such as Ev 2 R offer strong semantic grounding, they often carry a heavy computational burden; on the contrary, simpler overlapor one-to-one matching metrics are far less demanding, but frequently diverge from human judgments.In this paper, we introduce SemQA, a lightweight and accurate evidence scoring metric that combines transformer-based question scoring with bidirectional NLI entailment in answers.SemQA is then evaluated through correlation analysis with existing metrics, examination of representative examples, and human evaluations.