Dynamic knowledge graph for evolving worlds

Sonfack Sounchio, Serge

doi:10.1007/s13748-026-00432-0

Return

Dynamic knowledge graph for evolving worlds

2026-01-01

PRE

Sonfack Sounchio, Serge *

Djuikem, Clotilde

Kamsu-Foguem, Bernard

Geneste, Laurent

Coudert, Thierry

DOI：10.1007/s13748-026-00432-0

Original

Original request for help

Save

Abstract

En 中文

Graph representations are intuitive approaches used to represent knowledge about a world such that humans and systems can understand and use it to carry out tasks intelligently. These representations comprise nodes designating entities of this world and their edges materializing the relationships between them. Graph knowledge structures have been proven efficient in intelligent systems such as question-answering, search, or recommender systems. However, most worlds, being dynamic due to events, facts, pieces of knowledge, or interactions, cause their graph representation to evolve. These changes render classical graph representations inadequate for representing dynamic worlds because they become inconsistent or incorrect as these worlds change. This study elaborates on a generalized definition of dynamic knowledge graphs and provides properties for dynamic knowledge graphs (DKGs) representation of evolving worlds. It proposes a formal definition of dynamic knowledge graphs that integrates the dynamic characteristics and factors of evolving worlds into their structure, yielding faithful representations for consistent and growing knowledge. In addition, it extends query possibilities of DKGs across graph elements.

Keywords:

Dynamic knowledge graph

Knowledge graph

Knowledge representation

Evolving world