ArgStrength2021

We examine the notion of an argument in logic-based argumentation frameworks, arguing that some of the common requirements from arguments in this context, such as support minimality and consistency, may be waived and/or taken care of by appropriate attack rules. We then discuss various ways of handling this, as well as making preferences among arguments, by some straightforward linguistic extensions and revised attack rules.

Ethos, the character of the speaker, is a powerful tool used to influence others through communication. Together with logos (arguments) and pathos (emotions of the audience), they constitute the key elements of Aristotle's Rhetoric. This talk presents The New Ethos - a theory of ethotic structures which is empirically derived from and verified on real-life argument data and applied to several domains with the ultimate goal of implementing technologies such as ethos mining and ethos analytics.

Argumentation enables formalization of single agent and multi-agent (distributed) non-monotonic reasoning. In this talk I will review recent work on a new dialectical approach to structured argumentation (Dialectical ASPIC+) that, under the standard Dung semantics, is fully rational under resource bounds. The approach thus improves on ASPIC+ (a general framework that has been shown to provide argumentative characterisations of various non-monotonic logics), which is only partially rational, under the tacit assumption that agents are omniscient (i.e., they Einloggenhave unlimited resources). Moreover, Dialectical ASPIC+ accommodates uses of argument that more faithfully reflect real world uses of argument, and hence rationality is not conditional on the properties of the preference relation used to determine when one argument successfully challenges (defeats) another. The talk will conclude by briefly pointing to future development of the dialectical approach so as to accommodate two generalisations of Dung's semantics that have been applied to ASPIC+; one that enables argumentation-based reasoning about preferences over arguments, and the other that provides for more fine grained evaluation of the status of arguments via graded generalizations of Dung's semantics.

The talk is motivated by the need of accounting for the practical syllogism (familiar since Aristotle) as a piece of defeasible reasoning. To meet the need, the talk first refers to ranking theory as an account of defeasible descriptive reasoning. It then argues that two kinds of ought need to be distinguished, purely normative and fact-regarding obligations (in analogy to intrinsic and extrinsic utilities). It continues arguing that both kinds of ought can be iteratively revised and thus can and should be represented by ranking functions, too, just as iteratively revisable beliefs. Its central proposal will then be that the fact-regarding normative ranking function must be conceived as the sum of a purely normative ranking function and an epistemic ranking function (as suggested in qualitative decision theory). This will be illustrated with the examples already discussed. The talk will close with some remarks on the prospects of turning this into joint conditional epistemic-deontic logic.

In this talk, I will focus on the issue of automatically assessing the strength of natural language arguments, where strength is often associated to the quality of arguments and the trustworthiness of the person proposing them. I will investigate the role of machine learning and reasoning methods to tackle this issue, focusing on the real world scenarios of political debates and persuasive essays.