nenae18/life-advice-rag-dataset
收藏资源简介:
该数据集是一个用于RAG检索的chunk转换数据集,基于圣经、佛教经典/解说、论语和人文名言资料。它包含4,007个chunk,来自83个来源,总字符数为3,656,354,数据格式为JSONL。主要用途包括基于证据的人生建议RAG、宗教和哲学观点比较检索,以及演示用的问答。数据集文件包括chunks.jsonl(包含检索用的文本chunk)、dataset_manifest.json(数据集统计和来源列表)和sample_questions.json(演示用问题示例)。每个chunk包含chunk_id、source、category(如圣经/佛教/儒学·论语/名言/其他)、source_type(如pdf/txt/csv/zip)、loc(页码或位置)和text(用于检索的文本chunk)。数据集旨在回答如好妻子、好丈夫、困难应对、领导力、人际关系等人生建议问题,并支持结合BM25关键词检索和向量检索的RAG管道。使用注意事项强调需要确认原始资料的版权和许可,建议私有使用或限制共享。
This dataset is a chunk conversion dataset for RAG retrieval, based on materials including the Bible, Buddhist scriptures/commentaries, the Analects of Confucius, and humanistic quotes. It contains 4,007 chunks from 83 sources, with a total character count of 3,656,354, and is stored in JSONL format. Its main applications include evidence-based life advice RAG, comparative retrieval of religious and philosophical viewpoints, and demo-based question answering. The dataset files include chunks.jsonl (containing text chunks for retrieval), dataset_manifest.json (dataset statistics and source list), and sample_questions.json (demo question examples). Each chunk contains chunk_id, source, category (e.g., Bible/Buddhism/Confucianism·Analects/Quotes/Other), source_type (e.g., pdf/txt/csv/zip), loc (page number or location), and text (the text chunk for retrieval). This dataset is designed to answer life advice questions such as those about good wives, good husbands, coping with difficulties, leadership, interpersonal relationships, etc., and supports RAG pipelines combining BM25 keyword retrieval and vector retrieval. Usage notes emphasize the need to confirm the copyright and license of the original materials, and recommend private use or restricted sharing.





