Abstract: Next POI recommendation plays a crucial role in delivering personalized location-based services, but it faces significant challenges in capturing complex user behavior and adapting to ...

Abstract: The modality gap between vision and text embeddings in CLIP presents a significant challenge for zero-shot image captioning, limiting effective cross-modal representation. Traditional ...