Practices, opportunities and challenges in the fusion of knowledge graphs and large language models
Linyue Cai, Cihang Yu, Yu Kang, Fu Dong Yu, Heng Zhang, Yong Zhao
Abstract
The fusion of Knowledge Graphs (KGs) and Large Language Models (LLMs) leverages their complementary strengths to address limitations of both technologies. This paper explores integration practices, opportunities, and challenges, focusing on three strategies: KG-enhanced LLMs (KEL), LLM-enhanced KGs (LEK), and collaborative LLMs and KGs (LKC). The study reviews these methodologies, highlighting their potential to enhance knowledge representation, reasoning, and question answering. We comprehensively compile and categorize key challenges such as knowledge acquisition and real-time updates, providing valuable directions for future research. The paper also discusses emerging techniques and applications to advance the synergy between KGs and LLMs. Overall, this work offers a comprehensive overview of the current landscape and the transformative potential of KG-LLM fusion across various domains.