DuPLeR

DuPLeR: Dual-Path LLM Reasoning for Multimodal Few-Shot Knowledge Graph Completion

Researchers introduce DuPLeR, a dual-path LLM reasoning framework for multimodal few-shot knowledge graph completion (KGC). The system addresses inductive KGC under data-scarce conditions by combining multimodal LLM-derived type priors with factual support structures and a dual-pathway module that regulates message passing with query-relevant multimodal signals. Experiments across eight inductive variants of two multimodal KG benchmarks demonstrate robust performance in few-shot and zero-shot settings.