UzEDSA: Uzbek Emotion and Sentiment Analysis Dataset

Description

UzEDSA (Uzbek Emotion and Sentiment Analysis Dataset) is an open-access dataset designed for emotion detection and sentiment analysis tasks in the Uzbek language. It has been developed to support Natural Language Processing (NLP) research for low-resource languages, with a particular focus on Uzbek. The dataset consists of curated and preprocessed Uzbek-language texts collected from various sources. Each text instance is annotated with corresponding sentiment or emotion labels, enabling its use in supervised machine learning and deep learning tasks. The dataset is suitable for training, evaluation, and benchmarking of modern NLP and Large Language Models (LLMs).

Steps to reproduce

1. Download the dataset files (XLSX or TSV format) from the Mendeley Data repository. 2. Load the dataset using a data processing library such as pandas: import pandas as pd df = pd.read_csv("UzEDSA_dataset.tsv", sep="\t") 3. Inspect the dataset structure and fields: df.head() 4. Preprocess the text data if necessary (e.g., tokenization, normalization). 5. Use the dataset for training and evaluation of NLP models for tasks such as: - emotion classification (single-label) - multi-label emotion detection - sentiment classification - sarcasm detection 6. Split the dataset into training and test sets or apply cross-validation. 7. Train machine learning or deep learning models using the provided annotations.

Institutions

• Alisher Navo'i Tashkent State University of Uzbek Language and Literature

  Tashkent, Tashkent
• National University of Uzbekistan

  Tashkent, Tashkent

Categories

Licence