YAGO2s: Modular High-Quality Information Extraction with an Application to Flight Planning
Fabian M. Suchanek, Johannes Hoffart, Erdal Kuzey, Edwin Lewis-Kelham · 2013
Abstract: In this paper, wepresent YAGO2s, the new edition of the YAGO ontology [SKW07, HSBW12]. The software architecture has been refactored from scratch, yielding adesign that modularizes both code and data. This modularization enables us to add in new data sources more easily, while still maintaining the high accuracy and coherence of the ontology. Thus, we believe that YAGO2s occupies asweetspot between acentralized design and acompletely distributed design. In this demo, we present an application of this design to the task of planning a flight. Our proposed system finds flights between all airports close to the departure city to all airports close to the destination city. 1 Knowledge Base Construction In recent years, many projects have successfully created large-scale knowledge bases (KBs) in an automated fashion. The KBs contain millions of entities (such as rivers, universities, people, and movies), and millions of facts about them (such as who acted in which movie, which riverislocated in which country,etc.). There are several strategies to build such KBs. One strategy is to accumulate and reconcile as much data as possible.