# Neural End-to-End Learning for Computational Argumentation Mining

## TL;DR

This paper explores neural end-to-end methods for computational argumentation mining, comparing dependency parsing and sequence tagging approaches, and finds that local tagging models with multi-task learning outperform more complex dependency parsing models.

## Contribution

It demonstrates that simple BiLSTM-based sequence tagging models with multi-task learning are more effective for argumentation mining than dependency parsing approaches.

## Key findings

- BiLSTM-based tagging models perform robustly across scenarios.
- Multi-task learning improves argumentation mining performance.
- Dependency parsing models underperform compared to tagging models.

## Abstract

We investigate neural techniques for end-to-end computational argumentation mining (AM). We frame AM both as a token-based dependency parsing and as a token-based sequence tagging problem, including a multi-task learning setup. Contrary to models that operate on the argument component level, we find that framing AM as dependency parsing leads to subpar performance results. In contrast, less complex (local) tagging models based on BiLSTMs perform robustly across classification scenarios, being able to catch long-range dependencies inherent to the AM problem. Moreover, we find that jointly learning 'natural' subtasks, in a multi-task learning setup, improves performance.