ExplainabilityTransparencyTrustworthinesscs.AI
Knowledge Graph Embeddings and Explainable AI
Federico Bianchi, Gaetano Rossiello, Luca Costabello, Matteo Palmonari, Pasquale Minervini
April 30, 2020
Knowledge Graphs for eXplainable Artificial Intelligence
90 citations
5 influential
Key Takeaways
-
Knowledge graph embeddings convert complex relationships into mathematical vectors, making it easier for AI systems to understand and reason about connections between entities—this is a foundational technique now widely used in modern AI applications.
-
Explainability is a critical challenge for knowledge graph embeddings, as the paper identifies that predictions made by these systems require dedicated methods and models to explain why the AI reached a particular conclusion, addressing a key concern in responsible AI deployment.
-
Multiple approaches exist for representing knowledge in vector spaces, suggesting that there is no single "best" method—organizations implementing these systems should evaluate different approaches based on their specific needs and explainability requirements.
-
The field has established evaluation methods for knowledge graph embeddings, indicating that responsible AI practitioners have tools available to assess whether these systems are performing reliably before deployment in high-stakes applications.
Abstract
Knowledge graph embeddings are now a widely adopted approach to knowledge representation in which entities and relationships are embedded in vector spaces. In this chapter, we introduce the reader to the concept of knowledge graph embeddings by explaining what they are, how they can be generated and how they can be evaluated. We summarize the state-of-the-art in this field by describing the approaches that have been introduced to represent knowledge in the vector space. In relation to knowledge representation, we consider the problem of explainability, and discuss models and methods for explaining predictions obtained via knowledge graph embeddings.