Klasifikasi Komponen Argumen Secara Otomatis pada Dokumen Teks berbentuk Esai Argumentatif

TL;DR

This paper presents an automatic method for classifying argument components in argumentative essays, exploring feature sets including keywords, achieving over 72% accuracy, and analyzing the impact of keyword features.

Contribution

Introduces a new feature set with keyword lists for argument component classification and evaluates its effectiveness compared to existing features.

Findings

Keyword features do not significantly improve classification accuracy.

Achieved 72.45% accuracy in argument component classification.

Existing features are weak in distinguishing major claim from claim.

Abstract

By automatically recognize argument component, essay writers can do some inspections to texts that they have written. It will assist essay scoring process objectively and precisely because essay grader is able to see how well the argument components are constructed. Some reseachers have tried to do argument detection and classification along with its implementation in some domains. The common approach is by doing feature extraction to the text. Generally, the features are structural, lexical, syntactic, indicator, and contextual. In this research, we add new feature to the existing features. It adopts keywords list by Knott and Dale (1993). The experiment result shows the argument classification achieves 72.45% accuracy. Moreover, we still get the same accuracy without the keyword lists. This concludes that the keyword lists do not affect significantly to the features. All features are still weak to classify major claim and claim, so we need other features which are useful to differentiate those two kind of argument components.