Materialized Knowledge Bases from Commonsense Transformers
This work provides an off-the-shelf resource for researchers in natural language processing, enabling easier use and analysis of commonsense knowledge, though it is incremental as it builds on existing COMET methodology.
The paper addresses the lack of publicly available materialized commonsense knowledge bases generated from pre-trained language models, by creating such a resource and analyzing its precision and recall with concrete metrics.
Starting from the COMET methodology by Bosselut et al. (2019), generating commonsense knowledge directly from pre-trained language models has recently received significant attention. Surprisingly, up to now no materialized resource of commonsense knowledge generated this way is publicly available. This paper fills this gap, and uses the materialized resources to perform a detailed analysis of the potential of this approach in terms of precision and recall. Furthermore, we identify common problem cases, and outline use cases enabled by materialized resources. We posit that the availability of these resources is important for the advancement of the field, as it enables an off-the-shelf-use of the resulting knowledge, as well as further analyses on its strengths and weaknesses.