Dense-ATOMIC: Towards Densely-connected ATOMIC with High Knowledge Coverage and Massive Multi-hop Paths

  • 2023-05-28 12:56:59
  • Xiangqing Shen, Siwei Wu, Rui Xia
  • 0

Abstract

ATOMIC is a large-scale commonsense knowledge graph (CSKG) containingeveryday if-then knowledge triplets, i.e., {head event, relation, tail event}.The one-hop annotation manner made ATOMIC a set of independent bipartitegraphs, which ignored the numerous links between events in different bipartitegraphs and consequently caused shortages in knowledge coverage and multi-hoppaths. In this work, we aim to construct Dense-ATOMIC with high knowledgecoverage and massive multi-hop paths. The events in ATOMIC are normalized to aconsistent pattern at first. We then propose a CSKG completion method calledRel-CSKGC to predict the relation given the head event and the tail event of atriplet, and train a CSKG completion model based on existing triplets inATOMIC. We finally utilize the model to complete the missing links in ATOMICand accordingly construct Dense-ATOMIC. Both automatic and human evaluation onan annotated subgraph of ATOMIC demonstrate the advantage of Rel-CSKGC overstrong baselines. We further conduct extensive evaluations on Dense-ATOMIC interms of statistics, human evaluation, and simple downstream tasks, all provingDense-ATOMIC's advantages in Knowledge Coverage and Multi-hop Paths. Both thesource code of Rel-CSKGC and Dense-ATOMIC are publicly available onhttps://github.com/NUSTM/Dense-ATOMIC.

 

Quick Read (beta)

loading the full paper ...