Saima-Manzoor/code-switching-codesaviours-si26-saima

Code-Switching Urdu-English Dataset

Dataset Description

This dataset is created for Urdu-English code-switching language identification. It contains sentences that include Urdu words, English words, and a small number of mixed-language entries.

The dataset was prepared by collecting and organizing code-switched Urdu-English sentences. Each sentence was divided into individual words, and every word was assigned a language label. The data was then converted into a CSV file for further use in Natural Language Processing (NLP) tasks.

Dataset Format

The dataset contains the following columns:

• sentence: The complete Urdu-English code-switched sentence.
• word: An individual word extracted from the sentence.
• label: The language category assigned to that word.

Label Meanings

• URD: Urdu word
• ENG: English word
• MIX: Mixed or code-switched word

Data Collection and Preparation

The sentences were organized into a labeled dataset. Each sentence was processed at the word level, and words were paired with their corresponding language labels.

Python and Pandas were used to flatten the sentence-level data into word-level entries and export the final dataset as dataset.csv.

Dataset Statistics

• Total Sentences: 339
• Total Labeled Words: 2684
• Urdu Words (URD): 1510
• English Words (ENG): 1170
• Mixed Words (MIX): 4

Intended Use

This dataset can be used for code-switching research, language identification, NLP experiments, and machine learning tasks involving Urdu-English mixed text.

Created By

Saima Manzoor
Code Saviours ML/AI Internship – SI-26