In the last two decades, the theory of argumentation has become one of the core areas of research in the domain of artificial intelligence. These two decades have witnessed a voluminous amount of documentation related to argumentation that has flooded the field of artificial intelligence. Needless to say, this particular type of literary content has enriched the various facets of artificial intelligence in general. The theory of argumentation has also become a subject matter of applied AI course when we talk about the academic domain. In the industrial domain, this theory has been applied to study cognitive mechanisms in human-robot interactions.
Formulation of the theory
The classical treatment of the argumentation theory in a philosophical laboratory is restricted to the study of assertions and the manner of their proposition and discussion. In the present times, we broaden the formulation stages of the theory into four different dimensions. In the first stage, we define the discrete components of an argument and study their dependence and interaction. In the second stage, we identify the course of rules which are to be followed in the process of argumentation. The third stage is concerned with distinguishing valid and reliable arguments from invalid and unreliable ones. In the final stage, we formulate the conditions under which the discussion should be continued and the limitations under which it becomes redundant.
Framework of argumentation in AI
The framework of argumentation in artificial intelligence can be studied under three broad themes. In the first theme, we endeavor to study the influence of non-classical logic for the examination of argumentation in AI. Secondly, we study different types of cognitive models which have influenced interactions between man and the machine. Lastly, we arrive at a visual representation of argumentation so that it becomes suitable for the application of various models of artificial intelligence.
The dialogue process
The process of argumentation aims to decide on various claims in the presentation of the justification of an assertion. If we can quantify this information, a new sample space of data can be created. This sample of data becomes available for both logical reasoning processes and machine learning treatment. That said, the limitation to this kind of treatment is that the dialectical aspects of the argument are left unattended. The advantage of this kind of treatment needs to be mentioned as well. When we go for computational treatment, most of the ideas and facts that we build upon are derived from philosophical analysis. Alternatively, we can go for the diagrammatic treatment of this argumentative framework. In this type of treatment, we make use of Wigmore diagrams to represent a syntax of information in the form of logical flowcharts. This helps us to synthesize a working model for further analysis.
Future research in the theory of argumentation and its application in AI would come from an extension of semantics. This would be helpful to conceive chatbots with great cognitive skills.