masoudkaviani/persian_news_typos

Corrected News Articles Dataset

Overview:

This dataset contains pairs of raw text and their corrected versions, providing a valuable resource for tasks related to text correction, language modeling, and natural language processing (NLP). The dataset has been derived from various news articles and reports, focusing on correcting typographical errors, grammatical mistakes, and stylistic inconsistencies in the text. Each entry in the dataset consists of two fields: text (the original, unedited article or report) and corrected_text (the edited and corrected version of the same text).

Purpose:

The primary purpose of this dataset is to facilitate research and development in text correction systems, language models, and tasks such as:

• Automatic text proofreading and error correction.
• Training models for grammar correction and style improvement.
• Evaluating NLP systems on their ability to detect and fix errors in real-world written content.

Dataset Structure:

Each row in the dataset represents a single example with the following fields:

1. text: The original article or report that may contain errors (e.g., typos, grammatical issues, or stylistic inconsistencies).
2. corrected_text: The same article or report after corrections have been applied, ensuring that the text is grammatically correct, stylistically polished, and free of errors.

Source:

The dataset appears to originate from news articles and reports, including educational and institutional updates, as well as coverage of events such as conferences, public announcements, and cultural programs. The text often includes quotes from officials, descriptions of events, and commentary on academic and social topics.

Example:

Here is an example of a single entry from the dataset:

• text:
  "وزیز علوم درجمع استادان تمونه: سن بازنشستگی استادان نمونه به ۷۰ سال افزایش می‌یابد دانشگاه باید مهد چالشهای گفتمانی و خط دهنده و برنامه‌ریز جریانات سیاسی باشد وزیر علوم در جمع استادان نمونه کشور گفت: از استادان نمونه کشور انتظار می‌رود که رویکرد دانایی محوری و گفتمان علمی را یه عنوان یک بحث فرهنگی در دانشگاه‌ها توسعه و رونق بخشند."

• corrected_text:
  "وزیر علوم درجمع استادان نمونه: سن بازنشستگی استادان نمونه به ۷۰ سال افزایش می‌یابد دانشگاه باید مهد چالشهای گفتمانی و خط دهنده و برنامه‌ریز جریانات سیاسی باشد وزیر علوم در جمع استادان نمونه کشور گفت: از استادان نمونه کشور انتظار می‌رود که رویکرد دانایی محوری و گفتمان علمی را به عنوان یک بحث فرهنگی در دانشگاه‌ها توسعه و رونق بخشند."

(Note: While the text is primarily in Persian, the dataset can still be used for NLP tasks involving multilingual models or models trained on Persian text.)

Potential Use Cases:

1. Text Correction Systems:

   • Train models to detect and fix errors in text automatically.
   • Develop tools for grammar and style refinement.

2. Language Modeling:

   • Fine-tune pre-trained language models on this dataset to improve their understanding of error correction tasks.

3. Educational Applications:

   • Create tools for teaching grammar, spelling, and writing skills.

4. Multilingual Processing:

   • Use the dataset to train models for Persian text processing, as the dataset contains text in Persian.

Dataset Format:

The dataset is structured as a table, where each row corresponds to a pair of text and corrected_text. It can be loaded into Python as a Pandas DataFrame or a Hugging Face Dataset for further processing.

Size:

The size of the dataset (~11k) depends on the number of text-correction pairs included in the Excel file. Each row represents one example.

Data Quality:

The corrections in the corrected_text field have been manually applied, ensuring high-quality and accurate results. This makes the dataset suitable for supervised learning tasks where the corrected text serves as the ground truth.

Limitations:

1. The dataset may primarily focus on Persian text, which could limit its applicability to other languages.
2. The corrections might reflect specific stylistic preferences, which could introduce bias in models trained on this dataset.