Sanskrit Post-OCR text correction dataset
收藏arXiv2022-11-15 更新2024-06-21 收录
下载链接：
https://github.com/ayushbits/pe-ocr-sanskrit
下载链接
链接失效反馈官方服务：
资源简介：
Sanskrit Post-OCR文本校正数据集是由印度理工学院孟买分校的研究人员创建，旨在解决Sanskrit语言数字化资源稀缺的问题。该数据集包含来自30本不同书籍的约218,000条句子，总计150万个单词，涵盖天文学、医学和数学等多个领域。数据集的创建过程涉及从历史文档中提取文本，并通过手动验证确保质量。该数据集主要应用于提高Sanskrit OCR系统的准确性和可靠性，解决由于历史文档扫描质量、版面设计等因素导致的OCR错误问题。
Sanskrit Post-OCR Text Correction Dataset was developed by researchers at the Indian Institute of Technology Bombay to address the scarcity of digital resources for the Sanskrit language. This dataset contains approximately 218,000 sentences from 30 distinct books, totaling 1.5 million words, covering multiple fields such as astronomy, medicine and mathematics. The dataset creation process involved extracting text from historical documents and ensuring data quality through manual verification. It is primarily used to improve the accuracy and reliability of Sanskrit OCR systems, resolving OCR errors caused by factors like the scanning quality and layout design of historical documents.
提供机构：
印度理工学院孟买分校创建时间：
2022-11-15
