A neural network to classify metaphorical violence on cable news

TL;DR

This paper introduces a neural network system integrated with a web annotation tool to identify and classify metaphors in news corpora, enabling transfer learning across different metaphor classes.

Contribution

It presents a novel system that combines neural networks with user-driven annotation for metaphor detection and transfer learning in news text analysis.

Findings

System successfully suggests metaphor annotations during user annotation.

Transfer learning allows classification of metaphors across different categories.

Improves efficiency of metaphor annotation in large corpora.

Abstract

I present here an experimental system for identifying and annotating metaphor in corpora. It is designed to plug in to Metacorps, an experimental web app for annotating metaphor. As Metacorps users annotate metaphors, the system will use user annotations as training data. When the system is confident, it will suggest an identification and an annotation. Once approved by the user, this becomes more training data. This naturally allows for transfer learning, where the system can, with some known degree of reliability, classify one class of metaphor after only being trained on another class of metaphor. For example, in our metaphorical violence project, metaphors may be classified by the network they were observed on, the grammatical subject or object of the violence metaphor, or the violent word used (hit, attack, beat, etc.).