The Impact of Modeling Overall Argumentation with Tree Kernels

32736

Authors

nan

Abstract

AbstractSeveral approaches have been proposed to model either the explicit sequential structure of an argumentative text or its implicit hierarchical structure. So far, the adequacy of these models of overall argumentation remains unclear. This paper asks what type of structure is actually important to tackle downstream tasks in computational argumentation. We analyze patterns in the overall argumentation of texts from three corpora. Then, we adapt the idea of positional tree kernels in order to capture sequential and hierarchical argumentative structure together for the first time. In systematic experiments for three text classification tasks, we find strong evidence for the impact of both types of structure. Our results suggest that either of them is necessary while their combination may be beneficial.

Keywords

nan

Full Content

Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 2379–2389 Copenhagen, Denmark, September 7–11, 2017. c⃝2017 Association for Computational Linguistics The Impact of Modeling Overall Argumentation with Tree Kernels Henning Wachsmuth Bauhaus-Universität Weimar Faculty of Media, Webis Group henning.wachsmuth@uni-weimar.de Giovanni Da San Martino Qatar Computing Research Institute Arabic Language Technologies gmartino@hbku.edu.qa Dora Kiesel Bauhaus-Universität Weimar Faculty of Media, VR Group dora.kiesel@uni-weimar.de Benno Stein Bauhaus-Universität Weimar Faculty of Media, Webis Group benno.stein@uni-weimar.de Abstract Several approaches have been proposed to model either the explicit sequential struc- ture of an argumentative text or its implicit hierarchical structure. So far, the adequacy of these models of overall argumentation remains unclear. This paper asks what type of structure is actually important to tackle downstream tasks in computational argu- mentation. We analyze patterns in the over- all argumentation of texts from three cor- pora. Then, we adapt the idea of positional tree kernels in order to capture sequential and hierarchical argumentative structure to- gether for the first time. In systematic ex- periments for three text classification tasks, we find strong evidence for the impact of both types of structure. Our results suggest that either of them is necessary while their combination may be beneficial. 1 Introduction Argumentation theory has established a number of major argument models focusing on different aspects, such as the roles of an argument’s units (Toulmin, 1958), the inference scheme of an argu- ment (Walton et al., 2008), or the support and attack relations between arguments (Freeman, 2011). The common ground of these models is that they con- ceptualize an argument as a conclusion (in terms of a claim) inferred from a set of pro and con premises (reasons), which in turn may be the conclusions of other arguments. For the overall argumentation of a monological argumentative text such as the one in Figure 1(a), this results in an implicit hierarchical structure with the text’s main claim at the lowest depth. In addition, the text has an explicit linguistic structure that can be seen as a regulated sequence of speech acts (van Eemeren and Grootendorst, 2004). hierarchical structure (argumentative depth) main claim (depth 0) sequential structure (ordering in text) 1 2 3 4 5 1 0 2 con argument unit (attack of parent node) pro argument unit (support of parent node) [1] The death penalty is a legal means that as such is not practicable in Germany. [2] For one thing, inviolable human dignity is anchored in our constitution, [3] and furthermore no one may have the right to adjudicate upon the death of another human being. [4] Even if many people think that a murderer has already decided on the life or death of another person, [5] this is precisely the crime that we should not repay with the same. (b) (a) monological argumentative text Figure 1: (a) Example text with five argument units, taken from the Arg-Microtexts corpus introduced in Section 3. (b) Graph visualization of the sequential and hierarchical overall argumentation of the text. Figure 1(b) …

Content truncated to 500 words

Labeled Text

AbstractSeveral approaches have been proposed to model either the explicit sequential structure of an argumentative text or its implicit hierarchical structure. So far, the adequacy of these models of overall argumentation remains unclear. This paper asks what type of structure is actually important to tackle downstream tasks in computational argumentation. We analyze patterns in the overall argumentation of texts from three corpora. Then, we adapt the idea of positional tree kernels in order to capture sequential and hierarchical argumentative structure together for the first time. In systematic experiments for three text classification tasks, we find strong evidence for the impact of both types of structure. Our results suggest that either of them is necessary while their combination may be beneficial.

Publication Details

Paper ID:
32736
Year:
2017
Publication:
TIPSTER
Publication Type:
conf
DOI:
nan