Frame Shift Prediction

TL;DR

This paper introduces the Frame Shift Prediction task, using graph attention networks with auxiliary training to predict cross-linguistic frame shifts, facilitating multilingual FrameNet creation.

Contribution

It proposes a novel task and demonstrates that graph attention networks can effectively learn and predict frame shifts across languages.

Findings

Graph attention networks successfully predict frame shifts.

Auxiliary training improves prediction accuracy.

Enables automatic multilingual FrameNet annotation.

Abstract

Frame shift is a cross-linguistic phenomenon in translation which results in corresponding pairs of linguistic material evoking different frames. The ability to predict frame shifts enables automatic creation of multilingual FrameNets through annotation projection. Here, we propose the Frame Shift Prediction task and demonstrate that graph attention networks, combined with auxiliary training, can learn cross-linguistic frame-to-frame correspondence and predict frame shifts.