An Introduction to AnyBURL
| Autor: | Christian Meilicke, Melisachew Wudage Chekol, Daniel Ruffinelli, Heiner Stuckenschmidt |
|---|---|
| Rok vydání: | 2019 |
| Předmět: | |
| Zdroj: | KI 2019: Advances in Artificial Intelligence ISBN: 9783030301781 KI |
| Popis: | Current research on knowledge graph completion is often concerned with latent approaches that are based on the idea to embed a knowledge graph into a low dimensional vector space. At the same time symbolic approaches have attracted less attention [13]. However, such approaches have a big advantage: they yield an explanation in terms of the rules that trigger a prediction. In this paper we propose a bottom-up technique for efficiently learning logical rules from large knowledge graphs inspired by classic bottom-up rule learning approaches as Golem [8] and Aleph [10]. Our approach is called AnyBURL (Anytime Bottom-Up Rule Learning). We report on experiments where we evaluated AnyBURL on datasets that have been labelled as hard cases for simple (rule-based) approaches. Our approach performs as good as and sometimes better than most models that have been proposed recently. Moreover, the required resources in terms of memory and runtime are significantly smaller compared to latent approaches. This paper is an extended abstract of an IJCAI 2019 paper [6]. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|