CONTEXT
Knowledge Graphs (KGs) represent knowledge in the form of relations between entities, referred to as facts, typically grounded in formal ontological models. Such machine-readable formats enable AI systems to make decisions using clear and verifiable data. At the same time, Large Language Models (LLMs) have revolutionized the landscape of AI and are widely utilized in various NLP tasks.
02
CHALLENGE
Despite their remarkable performance, LLMs suffer from significant drawbacks. First, they are trained on general-purpose data and have lower performance in domain-specific tasks and low-resource languages. Secondly, they often reflect societal biases present in training data. Third, they sometimes produce inaccurate or made-up information, termed “hallucinations”. Finally, understanding the decision-making process of LLMs is challenging and their outputs may lack consistency.
03
SOLUTION
A potential solution to all these problems is to integrate LLMs with KGs, since KGs can provide factual information and the ability to perform reasoning. This would boost the LLM’s domain-specific reasoning, and interpretability, and mitigate biases and hallucinations. Conversely, since KGs require frequent updates by processing vast textual datasets, LLMs can aid in generating and refining them. Combining LLMs and KGs offers a promising opportunity to advance both technologies and represents a pivotal challenge in the contemporary research landscape.
Our Mission and Objectives
The main aim and objective of the KGELL Action is to develop and evaluate an automated pipeline that uses LLMs to generate and refine Knowledge Graph content from domain-specific textual datasets. The primary focus of this pipeline is reducing hallucinations and increasing factual accuracy.Specific Objectives
To achieve our main aim, the Action pursues specific objectives divided into two core areas.
Research Coordination:
Design a comprehensive research agenda to address current gaps and emerging trends in KGs and LLMs.
Bridge academia and industry by focusing on practical applications and challenges.
Foster interdisciplinary collaboration to develop shared evaluation standards in the field.
Promote knowledge sharing accessibility through a collaborative and open-access infrastructure.
Capacity Building:
Define and disseminate best practices for KGs and LLMs, providing practical guidance for researchers and practitioners.
Strengthen the research community through inclusive, ongoing engagement initiatives like conferences, seminars, and webinars.
Support the capacity development of early-career researchers through structured mentoring.
