publication venue for Verifying Cross-Modal Entity Consistency in News Using Vision-Language Models 2025