In Argument Retrieval, not only the relevance of an argumentative text with respect to the input query needs to be assessed. To truly fulfill a user's information need within the highly convoluted arena of argumentation, other qualitative aspects, such as the appropriateness and clarity of the argument play a crucial role. These concepts relate to the so-called "theory-based" notion of argument quality, which decomposes overall quality into a series of fine-grained qualitative aspects, each relating to either the logical, rhetorical, or dialectical dimension of arguments. However, despite the fact that this notion offers the potential for more advanced and targeted argument retrieval, the landscape of research efforts on theory-based argument quality in computational argumentation is still scarce. In this talk, we discuss the current state of research on theory-based argument quality in NLP, its opportunities for argument retrieval, as well as its current and future challenges.
