Maktabati/openiti-vectors
收藏资源简介:
该数据集包含完全向量化的OpenITI RELEASE 2025-1-9古典伊斯兰文本集合,专为语义搜索(RAG)准备。每个条目代表OpenITI语料库中8,943部作品之一的文本块,以及其嵌入向量和完整的元数据。统计信息:4,696,703个文本块;8,943部作品(仅主要版本,来自OpenITI TSV的status=pri);约470个Parquet文件(每个约45MB,zstd压缩);时间跨度:伊斯兰历1世纪至14世纪;语言:阿拉伯语(主要)、波斯语、土耳其语、乌尔都语。
This dataset contains the fully vectorized OpenITI RELEASE 2025-1-9 collection of classical Islamic texts, specifically tailored for semantic search (RAG) applications. Each entry corresponds to a text chunk from one of the 8,943 works in the OpenITI corpus, paired with its embedding vectors and complete metadata. Key statistical metrics: 4,696,703 text chunks; 8,943 primary-version works (filtered by the status=pri tag from the OpenITI TSV dataset); approximately 470 Parquet files, each ~45 MB in size and compressed with zstd; Time span: 1st century AH to 14th century AH; Languages: primarily Arabic, alongside Persian, Turkish, and Urdu.




