# MOROCO: The Moldavian and Romanian Dialectal Corpus

## TL;DR

This paper introduces MOROCO, a large dialectal corpus of Moldavian and Romanian texts, enabling diverse classification tasks and demonstrating approaches from string kernels to deep neural networks.

## Contribution

The creation of the MOROCO corpus with detailed dialectal and topical labels, and the evaluation of both shallow and deep models for dialect and topic classification.

## Key findings

- Deep models outperform shallow approaches.
- Dialectal and topical features are highly discriminative.
- Named entity removal affects model performance.

## Abstract

In this work, we introduce the MOldavian and ROmanian Dialectal COrpus (MOROCO), which is freely available for download at https://github.com/butnaruandrei/MOROCO. The corpus contains 33564 samples of text (with over 10 million tokens) collected from the news domain. The samples belong to one of the following six topics: culture, finance, politics, science, sports and tech. The data set is divided into 21719 samples for training, 5921 samples for validation and another 5924 samples for testing. For each sample, we provide corresponding dialectal and category labels. This allows us to perform empirical studies on several classification tasks such as (i) binary discrimination of Moldavian versus Romanian text samples, (ii) intra-dialect multi-class categorization by topic and (iii) cross-dialect multi-class categorization by topic. We perform experiments using a shallow approach based on string kernels, as well as a novel deep approach based on character-level convolutional neural networks containing Squeeze-and-Excitation blocks. We also present and analyze the most discriminative features of our best performing model, before and after named entity removal.