CSKG: The Commonsense Knowledge Graph

Commonsense reasoning is an important aspect of building robust AI systems and is receiving significant attention in the natural language understanding, computer vision, and knowledge graphs communities. At present, a number of valuable commonsense knowledge sources exist, with different foci, strengths and weaknesses. I will present an overview of existing sources of commonsense knowledge, including commonsense knowledge graphs, lexical resources, and visual commonsense sources. I will discuss how these can be harmonized in a single Commonsense Knowledge Graph (CSKG), organized into high-level dimensions of common sense, and enriched with logical axioms and preconditions. I will discuss our neuro-symbolic models that combine CSKG with neural language models to reason in question answering, natural language inference, and story understanding tasks. I will conclude with a list of remaining fundamental challenges for organizing commonsense knowledge and using it in combination with today’s neural techniques.