Can LLMs be Good Graph Judger for Knowledge Graph Construction?