An optimized Graph Embedding based Knowledge Graph Cleaning Algorithm

Data quality of knowledge graph is a one of the most important guareentees for many knowledge-based applications. We investigate the konwledge graph cleaning problem. We propose a knowledge graph error detection framework and design an optimized embedding based clean algorithm. The framework maps the knowledge graph into an numerical space and keeps the relationship between different nodes. With this framework, both miss data error and errous relationship can be cleaned. Extensive experimental study over different data sets validate the effectiveness of the method.