Argument Mining

8A Computational Model and a Simple Grammar-Based Implementation

The theoretical aspects of argumentation have been developed in Chapters 1 and 2. The linguistic dimensions of argumentation are presented in Chapters 3 and 4, where some of the main linguistic features of the structure of claims and justifications have been presented. Then, how the various parts of arguments can be annotated is discussed in Chapter 6. The main systems and technical challenges have been presented in Chapter 7.

Given these elements, this chapter introduces a way to model these observations in a computational way together with implementation elements, so that arguments can be mined in various types of texts. The approach presented here is based on grammars, a universal model that can be implemented in various ways. We show, in particular, how systematic observations of arguments and their annotations can progressively be modeled by means of grammars. The implementation presented here is simple: our goal is to show a possible implementation of the linguistic principles developed in the previous chapters and how the data identified via annotation as typical argumentation can be used.

This chapter deals first with the identification of argumentative units, which is a very difficult problem, but a preliminary step toward the other processes. Then, lexical and grammatical formalisms and data used in structure analysis are developed so that the reader may see what features are necessary and how they can ...