Abstract
Despite extensive research in the past years, the computational modeling of argumentation remains challenging. The primary reason lies in the inherent complexity of the human processes behind, which commonly requires the integration of extensive knowledge far beyond what is needed for many other natural language understanding tasks. Existing work on the mining, assessment, reasoning, and generation of arguments acknowledges this issue, calling for more research on the integration of common sense and world knowledge into computational models. However, a systematic effort to collect and organize the types of knowledge needed is still missing, hindering targeted progress in the field. In this opinionated survey paper, we address the issue by (1) proposing a pyramid of types of knowledge required in computational argumentation, (2) briefly discussing the state of the art on the role and integration of these types in the field, and (3) outlining the main challenges for future work.
Abstract (translated)
URL
https://arxiv.org/abs/2107.00281
