Predicting the Co-Evolution of Event and Knowledge Graphs
This addresses the challenge of dynamic knowledge graphs for applications requiring real-time updates, though it is incremental by extending static models to time-dependent cases.
The paper tackles the problem of modeling time-dependent knowledge graphs by predicting future events that cause changes, using both background knowledge and recent events. The approach performs well in clinical, recommendation, and sensor network applications.
Embedding learning, a.k.a. representation learning, has been shown to be able to model large-scale semantic knowledge graphs. A key concept is a mapping of the knowledge graph to a tensor representation whose entries are predicted by models using latent representations of generalized entities. Knowledge graphs are typically treated as static: A knowledge graph grows more links when more facts become available but the ground truth values associated with links is considered time invariant. In this paper we address the issue of knowledge graphs where triple states depend on time. We assume that changes in the knowledge graph always arrive in form of events, in the sense that the events are the gateway to the knowledge graph. We train an event prediction model which uses both knowledge graph background information and information on recent events. By predicting future events, we also predict likely changes in the knowledge graph and thus obtain a model for the evolution of the knowledge graph as well. Our experiments demonstrate that our approach performs well in a clinical application, a recommendation engine and a sensor network application.