The Sequential Fusion technique, presented in the work by [65], is a two-phase method designed to improve domain-specific LLMs by integrating information from complex settings. In the first phase, general LLMs build Knowledge Graphs (KGs) from complex texts using a relation extraction procedure guided by prompt modules that provide reasoning processes, output formats, and guidelines to minimize ambiguity. In the second phase, a Structured Knowledge Transformation (SKT) module converts the structured knowledge from the KGs into natural language descriptions, which are then used to update domain-specific LLMs via the Knowledge Editing (IKE) method without requiring significant retraining.

Authors

• Person: Not available Organization: Springer
  A survey on augmenting knowledge graphs (KGs) with large ...

Sources

• A survey on augmenting knowledge graphs (KGs) with large ... link.springer.com Springer via serper

Referenced by nodes (2)

• Large Language Models concept
• knowledge graphs concept