InfoBayAI/Marathi-Non-STEM-Educational-Text-Corpus
收藏资源简介:
该数据集是一个大规模的马拉地语非STEM教科书数据集合,包含584本书籍和3736万单词,旨在支持马拉地语语言理解、推理和常识学习的高级NLP系统和AI模型的开发和训练。作为大规模多语言教育语料库的一部分,它涵盖5000多个主题,超过26亿单词,并包含交织的图像以增强上下文理解,总计包括15种语言的39000多本书籍,专注于多语言知识学习、推理和教育理解。数据集规格包括:书籍数量584本,单词数3736万,模态为马拉地语,类型为教育/非STEM,数据来源为精选的学术教科书和教育材料,数据性质为真实世界和精选数据,内容涵盖人文、社会科学、文学、历史、地理、公民教育和教育解释。关键用例包括非STEM内容中的命名实体识别、教育文本摘要和理解、自动辅导和教育助手、知识检索系统以及模型评估和基准测试。其价值在于支持马拉地语非STEM学科学习、提升AI模型的推理和理解能力、支持多语言和领域特定NLP系统、帮助构建AI驱动的教育平台,并增强LLM在非STEM领域的准确性和可靠性。
This dataset is a large-scale collection of Marathi Non-STEM textbook data, containing 584 books and 37.36 million words, designed to support the development and training of advanced NLP systems and AI models for language understanding, reasoning, and general knowledge learning in Marathi. It is part of a large-scale multilingual educational corpus containing over 2.6+ billion words across 5,000+ subjects, supported by interwoven images for deeper contextual understanding, and includes 39,000+ books across 15 languages, designed to support the development and training of advanced NLP systems and AI models for multilingual knowledge learning, reasoning, and educational understanding. Dataset specifications include: Books: 584, words: 37.36 million, Modality: Marathi, Type: Educational / Non-STEM, Data Source: Curated academic textbooks and educational material, Data Nature: Real-world and curated data, Content: Humanities, social sciences, literature, history, geography, civics, and educational explanations. Key use cases include Named Entity Recognition (NER) in Non-STEM content, educational text summarization and comprehension, automated tutoring and educational assistants, knowledge retrieval systems, and model evaluation and benchmarking. Its value enables learning of Non-STEM subjects in Marathi, improves reasoning and comprehension capabilities in AI models, supports multilingual and domain-specific NLP systems, helps build AI-powered educational platforms, and enhances accuracy and reliability of LLMs in Non-STEM domains.




