This paper explores the integration of knowledge graphs (KGs) and large language models (LLMs) to enhance AI applications. It reviews 28 papers published within five years, analyzing methods for KG-powered LLMs, LLM-based KGs, and hybrid approaches. The study highlights key trends, innovative techniques, and common challenges in combining KGs and LLMs. It emphasizes the complementary strengths of KGs in providing structured knowledge and LLMs in generating text and understanding language. KGs can improve LLMs by providing factual grounding, enhancing interpretability, and supporting semantic understanding. Conversely, LLMs can assist in KG construction by automating information extraction and improving the efficiency of knowledge graph building. The paper also discusses the benefits of hybrid approaches that combine both technologies, leading to better performance in tasks like named entity recognition and relation classification. However, challenges such as the need for frequent updates to KGs and LLMs, computational costs, and the difficulty of integrating knowledge from KGs into LLMs remain. The study concludes that while combining KGs and LLMs offers significant advantages, further research is needed to address current limitations and improve knowledge integration methods. The integration of KGs and LLMs represents a critical trend in AI, with potential to create more reliable and context-aware systems.This paper explores the integration of knowledge graphs (KGs) and large language models (LLMs) to enhance AI applications. It reviews 28 papers published within five years, analyzing methods for KG-powered LLMs, LLM-based KGs, and hybrid approaches. The study highlights key trends, innovative techniques, and common challenges in combining KGs and LLMs. It emphasizes the complementary strengths of KGs in providing structured knowledge and LLMs in generating text and understanding language. KGs can improve LLMs by providing factual grounding, enhancing interpretability, and supporting semantic understanding. Conversely, LLMs can assist in KG construction by automating information extraction and improving the efficiency of knowledge graph building. The paper also discusses the benefits of hybrid approaches that combine both technologies, leading to better performance in tasks like named entity recognition and relation classification. However, challenges such as the need for frequent updates to KGs and LLMs, computational costs, and the difficulty of integrating knowledge from KGs into LLMs remain. The study concludes that while combining KGs and LLMs offers significant advantages, further research is needed to address current limitations and improve knowledge integration methods. The integration of KGs and LLMs represents a critical trend in AI, with potential to create more reliable and context-aware systems.