The field of knowledge graph completion is moving towards more advanced and innovative methods to improve the accuracy and expressivity of models. Researchers are exploring new approaches to address the limitations of traditional methods, such as rank bottlenecks and static scoring functions. The use of context-aware and semantic-aware methods is becoming increasingly popular, allowing for more nuanced and informed predictions. Additionally, there is a growing interest in temporal knowledge graph reasoning, which aims to model the dynamic development of facts over time. Notable papers in this area include:

• A paper proposing KGE-MoS, a mixture-based output layer to break rank bottlenecks in knowledge graph completion models, which shows significant improvements in performance and probabilistic fit.
• A paper introducing Flow-Modulated Scoring, a framework that combines context-sensitive entity representations with dynamic transformation to model relational semantics, achieving state-of-the-art results on several benchmarks.