Learning to Give a Complete Argument with a Conversational Agent: An Experimental Study in Two Domains of Argumentation

This data is collected to find out how having a conversation with our agent affects argumentation. To model the arguments, we used Toulmin's model of argument. Based on the model, a good argument contains 3 different parts: 1. Claim, 2. Warrant, 3. Evidence. Based on Toulmin's model, these three components are the core components of arguments. This dataset has been collected during a between-subject experiment in which the treatment groups first talked to our agent in Task 1 and received feedback on faulty structural arguments and then did Task 2 and 3 which were answering a question on the same and completely different topic in comparison to Task 1.