Type-augmented Relation Prediction in Knowledge Graphs

Knowledge graphs (KGs) are of great importance to many real world applications, but they generally suffer from incomplete information in the form missing relations between entities. graph completion (also known as relation prediction) is task inferring facts given existing ones. Most work proposed by maximizing likelihood observed instance-level triples. Not much attention, however, paid ontological information, such type entities and relations. In this work, we propose a type-augmented prediction (TaRP) method, where apply both for prediction. particular, encoded prior probabilities likelihoods respectively, combined following Bayes' rule. Our TaRP method achieves significantly better performance than state-of-the-art methods on four benchmark datasets: FB15K, FB15K-237, YAGO26K-906, DB111K-174. addition, show that improved data efficiency. More importantly, extracted specific dataset can generalize well different datasets through model.