The field of natural language processing and knowledge graph research is moving towards tighter integration of knowledge graphs and large language models. Recent studies have shown that incorporating knowledge graphs into large language models can significantly improve their performance on various tasks, including question answering, text generation, and decision-making. The use of knowledge graphs enables the models to capture complex relationships between entities and concepts, leading to more accurate and informative outputs. Furthermore, the development of novel frameworks and architectures, such as multimodal graph assistants and graph-retrieval-augmented generation, has facilitated more efficient and effective integration of knowledge graphs and large language models. Notable papers in this area include STORYTELLER, which introduces a novel plot-planning framework for coherent and cohesive story generation, and MLaGA, which proposes a multimodal large language and graph assistant for reasoning over complex graph structures and multimodal attributes. Additionally, papers like E^2GraphRAG and GraphRAG-Bench have made significant contributions to the development of more efficient and effective graph-based retrieval-augmented generation methods.
Advancements in Integrating Knowledge Graphs and Large Language Models
Sources
E^2GraphRAG: Streamlining Graph-based RAG for High Efficiency and Effectiveness
Leveraging Knowledge Graphs and LLMs for Structured Generation of Misinformation
Guiding Generative Storytelling with Knowledge Graphs
STORYTELLER: An Enhanced Plot-Planning Framework for Coherent and Cohesive Story Generation
GraphRAG-Bench: Challenging Domain-Specific Reasoning for Evaluating Graph Retrieval-Augmented Generation
MLaGA: Multimodal Large Language and Graph Assistant
Knowledge Graph Completion by Intermediate Variables Regularization
Dynamic Programming Techniques for Enhancing Cognitive Representation in Knowledge Tracing
Entity-Augmented Neuroscience Knowledge Retrieval Using Ontology and Semantic Understanding Capability of LLM
Human-In-The-Loop Workflow for Neuro- Symbolic Scholarly Knowledge Organization
KG-BiLM: Knowledge Graph Embedding via Bidirectional Language Models
Signals as a First-Class Citizen When Querying Knowledge Graphs
Graph-Embedding Empowered Entity Retrieval
Graph Counselor: Adaptive Graph Exploration via Multi-Agent Synergy to Enhance LLM Reasoning
A Generative Adaptive Replay Continual Learning Model for Temporal Knowledge Graph Reasoning
A Graph-Retrieval-Augmented Generation Framework Enhances Decision-Making in the Circular Economy
Evaluating the Effectiveness of Linguistic Knowledge in Pretrained Language Models: A Case Study of Universal Dependencies