Upon successful completion of this course, the student:
- knows the essence of formal models for argumentation as reasoning and as dialogue;
- is able to apply these formalisms to formal examples;
- has insight into the metatheoretic properties of the studied formalisms and can verify simple metatheoretic properties;
- is aware of the main formal relations between different models of argumentation, and can formally verify simple relations;
- is able to evaluate the suitability of the studied formalisms for modeling realistic forms of argumentation;
- is able to model realistic reasoning problems in an implemented argumentation reasoner.
Grading
A written midterm and final exam (40% each) and a homework assignment to be done in pairs (20%). To pass the course, the minimum grade for each of these tests must be at least 4.0.
The repair test requires that:
- the original final grade is at least a 4 or an AANV, and
- the student has handed in a version of the homework exercise that at least satisfies the minimum requirements for this assignment.
Prerequisites
At the start of the course the student should have a knowledge of standard propositional and first-order predicate logic and elementary set theory as taught in INFOB1L1 Logica voor Informatica, or an equivalent course
|
 |
|
This course gives an introduction to the computational study of argumentation in AI, a currently popular subfield of symbolic AI. The course especially focuses on formal models of argumentation and their application in areas like commonsense reasoning, legal reasoning and multi-agent interaction.
The computational study of argumentation concerns two aspects: reasoning and dialogue. Argumentation as a form of reasoning makes explicit the reasons for the conclusions that are drawn and how conflicts between reasons are resolved. Systems for argumentation-based inference were originally developed in the field of nonmonotonic logic, which formalizes qualitative reasoning with incomplete, uncertain or inconsistent information. Argument-based systems have been very successful as nonmonotonic logics, since they are based on very natural concepts, such as argument, counterargument, rebuttal and defeat. In this course the following formalism will be discussed:
- sefault logic (a still influential early nonmonotonic logic)
- the theory of abstract argumentation frameworks (the generally accepted formal foundation of the field)
- the theory of structured argumentation frameworks, with a special focus on the ASPIC+ approach.
- formal accounts of change operations on argumentation frameworks
- formal models of legal case-based reasoning
Argumentation as a form of dialogue concerns the rational resolution of conflicts of opinion by verbal means. Intelligent agents may disagree, for instance, about the pros and cons of alternative proposals, or about the factual basis of such proposals. Dialogue systems for argumentation formally define protocols for argumentation dialogues and thus enable a formal study of the dynamics of argumentative agent interaction, including issues of strategic choice. In this course two examples of such dialogeus systems will be discussed.
This course is the replacement of INFOCR Commonsense Reasoning and Argumentation. Only one of the two courses can be part of your graduation program.
Course form
Interactive lectures (14x2 hours) plus self-study with exercises.
Literature
Online reader, online articles and educational software tools
|
|
|