Knowledge Graphs (KGs) are used to organize and understand vast amounts of information. However, they usually suffer from incompleteness which limits their applicability. Knowledge Graph Completion (KGC) is a critical process for augmenting these graphs and is typically achieved through embedding-based and pre-trained language model based methods. Although Large Language Models (LLMs) have demonstrated significant potential for knowledge extraction and reasoning, they face challenges such as fact knowledge hallucinations which adversely affect their KGC performance. We propose Knit, a novel KG-integrated instruction tuning framework to alleviate LLM fact knowledge Hallucinations in KGC. The proposed framework comprises three key components: (1) KG-integrated information adapter, (2) knowledge prompts, and (3) KG-integrated instruction tuning strategy. These components enhance LLMs’ ability to recognize entity relationships, improve embedding interpretability, and ensure response consistency. The experimental results across four public datasets (WN11, FB13, WN18RR, and YAGO3-10) demonstrate that Knit achieves state-of-the-art performance in KGC with significant improvements in all KGC subtasks.

Emergency navigation with a large number of sensors can serve as a safety service in emergencies. Recent studies have focused on navigation protocols to safely guide people to exits while helping them avoid hazardous areas. However, those approaches are not applicable in all circumstances. Both the dynamics of the environment and the mobility of users are key challenges for the efficiency and effectiveness of navigation protocols. The concepts of navigability and reachability are used to evaluate three typical navigation approaches. A large number of simulation results show that these two indicators effectively identify the performance levels of navigation protocols in changing environments.