This paper presents a parallel corpus of raw texts between the Uzbek and Kazakh languages as a dataset for machine translation applications, focusing on the data collection process, dataset description, and its potential for reuse. The dataset-building process includes three separate stages, starting with a tiny portion of already available parallel data, then some more compiled from openly available resources like literature books, and web news texts, which were aligned using the sentence alignment method, encompassing a wide range of topics and genres. Finally, the majority of the dataset was taken from a raw text corpus in Uzbek and manually translated into Kazakh by a group of experts who are fluent in both languages. The resulting parallel corpus serves as a valuable resource for researchers and practitioners interested in Kazakh and Uzbek language processing tasks, particularly in the context of neural machine translation, where the presented data can be used for testing the rule-based machine translation models, or it can be used for both training statistical and neural machine translation models as well. The dataset has been made accessible through the widely recognized Hugging Face platform, a repository known for facilitating collaborative efforts and advancing Natural Language Processing (NLP) applications. This combination of methods to obtain a parallel corpus plays as a pivot for other languages among other low-resource Turkic languages.

Making natural language processing technologies available for low-resource languages is an important goal to improve the access to technology in their communities of speakers. In this paper, we provide the first annotated corpora for polarity classification for Uzbek language. Our methodology considers collecting a medium-size manually annotated dataset and a larger-size dataset automatically translated from existing resources. Then, we use these datasets to train sentiment analysis models on the Uzbek language, using both traditional machine learning techniques and recent deep learning models.

In this paper, we introduce UzbekPOS — a part-of-speech (POS) tagged dataset manually annotated for the Uzbek language, designed for natural language processing, artificial intelligence models, and corpus linguistics applications. This tagged corpus is currently the largest publicly available POS-tagged corpus for the Uzbek language. The dataset comprises sentences drawn from a diverse range of Uzbek text sources, including literature, news outlets, science, education, and public speaking, to reflect linguistic and topical diversity. The sentences are tokenized and annotated by professional annotators, utilizing a finely grained POS tagset which integrates standard Universal Dependencies with additional labels that are specific to the morphological and syntactic features of the Uzbek language, comprising 16 tags in total.

There has been an increasing interest in learning cross-lingual word embeddings to transfer knowledge obtained from a resource-rich language, such as English, to lower-resource languages for which annotated data is scarce, such as Turkish, Russian, and many others. In this paper, we present the first viability study of established techniques to align monolingual embedding spaces for Turkish, Uzbek, Azeri, Kazakh and Kyrgyz, members of the Turkic family which is heavily affected by the low-resource constraint. Those techniques are known to require little explicit supervision, mainly in the form of bilingual dictionaries, hence being easily adaptable to different domains, including low-resource ones. We obtain new bilingual dictionaries and new word embeddings for these languages and show the steps for obtaining cross-lingual word embeddings using state-of-the-art techniques. Then, we evaluate the results using the bilingual dictionary induction task. Our experiments confirm that the obtained bilingual dictionaries outperform previously-available ones, and that word embeddings from a low-resource language can benefit from resource-rich closely-related languages when they are aligned together. Furthermore, evaluation on an extrinsic task (Sentiment analysis on Uzbek) proves that monolingual word embeddings can, although slightly, benefit from cross-lingual alignments.