Large Language Models in Knowledge Graphs and Entity Structure Discovery

The field of knowledge graphs and entity structure discovery is witnessing significant advancements with the integration of large language models (LLMs). Researchers are leveraging LLMs to improve the accuracy and efficiency of tasks such as instance completion, schema mapping, and entity structure extraction. A notable trend is the use of LLMs to automate tasks that were previously manual and time-consuming, such as schema generation and competency question generation. Additionally, LLMs are being used to enhance the performance of zero-shot event detection and open-schema entity structure discovery. These advancements have the potential to revolutionize the way knowledge graphs are constructed and utilized. Noteworthy papers include GenIC, which proposes a two-step generative instance completion framework, and Zero-Shot Open-Schema Entity Structure Discovery, which introduces a novel approach to entity structure extraction without requiring predefined schemas or annotated samples. Bench4KE is also notable for establishing a benchmarking system for evaluating competency question generation tools.

Sources

GenIC: An LLM-Based Framework for Instance Completion in Knowledge Graphs

Bench4KE: Benchmarking Automated Competency Question Generation

Towards Scalable Schema Mapping using Large Language Models

Zero-Shot Open-Schema Entity Structure Discovery

Schema Generation for Large Knowledge Graphs Using Large Language Models

DiCoRe: Enhancing Zero-shot Event Detection via Divergent-Convergent LLM Reasoning

Built with on top of