Entity Alignment in Multi-lingual, Temporal, and Probabilistic Knowledge Graphs

Abstract

This Thesis focuses on dynamic knowledge graphs, specifically addressing entity alignment. In temporal scenarios, take weather forecasting as an example. Our methods can analyze historical data from knowledge graphs over time, improving prediction accuracy. In probabilistic knowledge graphs, in medical research, it deals with uncertain patient symptoms and test results. Doctors can make more informed diagnoses, potentially saving lives. By streamlining knowledge utilization, it not only boosts efficiency in these diverse fields but also enriches user experiences, ultimately bringing widespread benefits to society, from enhancing scientific research capabilities to making daily life more convenient.