UWV/wim-instruct-signaalberichten-to-jsonld-agent-steps
收藏Hugging Face2025-07-23 更新2025-10-18 收录
下载链接：
https://hf-mirror.com/datasets/UWV/wim-instruct-signaalberichten-to-jsonld-agent-steps
下载链接
链接失效反馈官方服务：
资源简介：
该数据集包含116,056个指令遵循示例，来自一个将荷兰市政投诉信息转换为结构化JSON-LD知识图的生产管道。每个示例代表一个原子LLM调用，用于从公民与政府通信中提取语义信息。数据集适用于荷兰语NLP任务，包括实体提取、分类、Schema.org映射、JSON-LD生成和主题分类。
This dataset contains 116,056 instruction-following examples from a production pipeline that converts Dutch municipality complaint messages into structured JSON-LD knowledge graphs. Each example represents an atomic LLM call for extracting semantic information from citizen-government communications. The dataset is suitable for Dutch NLP tasks including entity extraction, classification, Schema.org mapping, JSON-LD generation, and topic classification.
提供机构：
UWV
