johneze/ktt-t22-grants-tenders

可重复的非洲联盟和地区资助/招标文件的合成数据集，旨在测试多语言信息检索和摘要系统。数据集包含32份英文招标文档（24份纯文本+8份HTML）、8份法文招标文档、10份商业配置文件（来自卢旺达、肯尼亚、塞内加尔、刚果民主共和国、埃塞俄比亚）以及30份专家策划的（配置文件，招标）黄金匹配对。每个招标文档包含标题、部门、预算、截止日期、地区、资格要求等字段。商业配置文件则包含ID、部门、国家、员工人数、语言、需求文本和过去资助等信息。

Reproducible synthetic dataset of African Union and regional grant/tender documents, designed to test multilingual information retrieval and summarization systems. The dataset includes 32 English tender documents (24 plain text + 8 HTML), 8 French tender documents, 10 business profiles (from Rwanda, Kenya, Senegal, DRC, Ethiopia), and 30 expert-curated (profile, tender) gold match pairs. Each tender document contains fields such as title, sector, budget, deadline, region, eligibility, etc. Business profiles include ID, sector, country, number of employees, languages, needs text, and past funding information.