Spoken dialect identification in Twitter using a multi-filter architecture

TL;DR

This paper introduces a multi-filter neural architecture for Swiss German dialect identification on Twitter, achieving high accuracy by combining recall-focused and precision-focused classifiers, specifically tailored for dialect detection.

Contribution

The novel multi-filter neural model effectively distinguishes Swiss German dialects from non-German tweets, improving dialect identification accuracy without using generic language identification.

Findings

Achieved F1-score of 0.982 on shared task test set

Utilized a multi-class classifier with GSW as one label

Designed filters to optimize recall and precision separately

Abstract

This paper presents our approach for SwissText & KONVENS 2020 shared task 2, which is a multi-stage neural model for Swiss German (GSW) identification on Twitter. Our model outputs either GSW or non-GSW and is not meant to be used as a generic language identifier. Our architecture consists of two independent filters where the first one favors recall, and the second one filter favors precision (both towards GSW). Moreover, we do not use binary models (GSW vs. not-GSW) in our filters but rather a multi-class classifier with GSW being one of the possible labels. Our model reaches F1-score of 0.982 on the test set of the shared task.