Russian Texts Detoxification with Levenshtein Editing

TL;DR

This paper presents a two-step tagging-based model for detoxifying toxic Russian texts by creating neutral versions, achieving state-of-the-art style transfer accuracy in a shared task.

Contribution

The paper introduces a novel tagging-based detoxification approach that outperforms larger sequence-to-sequence models on Russian texts.

Findings

Achieved highest style transfer accuracy in RUSSE Detox shared task.

Surpassed larger sequence-to-sequence models in detoxification quality.

Demonstrated effectiveness of a two-step tagging approach for text detoxification.

Abstract

Text detoxification is a style transfer task of creating neutral versions of toxic texts. In this paper, we use the concept of text editing to build a two-step tagging-based detoxification model using a parallel corpus of Russian texts. With this model, we achieved the best style transfer accuracy among all models in the RUSSE Detox shared task, surpassing larger sequence-to-sequence models.