Ascent KB

홈페이지 · 논문 9편

This dataset contains 8.9M commonsense assertions extracted by the Ascent pipeline developed at the Max Planck Institute for Informatics. The focus of this dataset is on everyday concepts such as *elephant*, *car*, *laptop*, etc. The current version of Ascent KB (v1.0.0) is approximately 19 times larger than ConceptNet (note that, in this comparison, non-commonsense knowledge in ConceptNet such as lexical relations is excluded).