Neurosymbolic Methods for Dynamic Knowledge Graphs
Mehwish Alam, Genet Asefa Gesese, Pierre-Henri Paris
TL;DR
This work formalizes dynamic knowledge graphs (DKGs) and surveys neurosymbolic learning approaches for dynamic settings, focusing on temporal and non-temporal KG completion and dynamic entity alignment. It classifies temporal knowledge graph completion (TKGC) methods, outlines training objectives and evaluation protocols, and surveys benchmark datasets; it also covers online and continual learning strategies (e.g., LKGE, IncLoRA) for incremental KG embedding. The chapter situates its contribution as a taxonomy of DKG representations and a synthesis of neurosymbolic techniques that preserve temporality and support evolving data, with discussion of challenges in scale, ontology use, and literal information. It highlights representative KGs (DBpedia, YAGO, Wikidata, EventGraph) and practical considerations for temporal encoding (Time Ontology, RDF-star, reification, named graphs) and for downstream tasks like TKGC and dynamic entity alignment. The work emphasizes future directions such as better integration of literals and LLMs, domain adaptation, and scalable, provenance-aware representations for billions-to-trillions of triples.
Abstract
Knowledge graphs (KGs) have recently been used for many tools and applications, making them rich resources in structured format. However, in the real world, KGs grow due to the additions of new knowledge in the form of entities and relations, making these KGs dynamic. This chapter formally defines several types of dynamic KGs and summarizes how these KGs can be represented. Additionally, many neurosymbolic methods have been proposed for learning representations over static KGs for several tasks such as KG completion and entity alignment. This chapter further focuses on neurosymbolic methods for dynamic KGs with or without temporal information. More specifically, it provides an insight into neurosymbolic methods for dynamic (temporal or non-temporal) KG completion and entity alignment tasks. It further discusses the challenges of current approaches and provides some future directions.