Neural String Edit Distance

TL;DR

The paper introduces a neural string edit distance model that improves string-pair matching and transduction by integrating learnable, differentiable edit distance into neural networks, balancing performance and interpretability.

Contribution

It modifies the expectation-maximization algorithm into a differentiable loss, enabling neural integration for enhanced string matching and transduction tasks.

Findings

Contextual representations achieve state-of-the-art performance.

Static embeddings offer interpretability with some accuracy trade-off.

Framework is versatile for various string-related tasks.

Abstract

We propose the neural string edit distance model for string-pair matching and string transduction based on learnable string edit distance. We modify the original expectation-maximization learned edit distance algorithm into a differentiable loss function, allowing us to integrate it into a neural network providing a contextual representation of the input. We evaluate on cognate detection, transliteration, and grapheme-to-phoneme conversion, and show that we can trade off between performance and interpretability in a single framework. Using contextual representations, which are difficult to interpret, we match the performance of state-of-the-art string-pair matching models. Using static embeddings and a slightly different loss function, we force interpretability, at the expense of an accuracy drop.