From One Point to A Manifold: Orbit Models for Knowledge Graph Embedding

Knowledge graph embedding aims at offering a numerical representation paradigm for knowledge by transforming the entities and relations into continuous vector space. This paper studies the problem of unsatisfactory precise prediction, that existing methods could not express the knowledge in a fine degree to make a precise prediction. To alleviate this issue, we propose an orbit-based embedding model (OrbitE). which is a well-posed algebraic system that expands the position of golden triples from one point in current models to a manifold. Extensive experiments show that the proposed model achieves substantial improvements against the state-of-the-art baselines especially for precise prediction task, with almost the fastest speed.