遇见数据集

LLM-Classified European Public IT Procurement Awards: Digitization, Digitalization, and Digital Transformation in German TED Notices, 2024–2025

收藏
Zenodo2026-05-14 更新2026-05-26 收录
官方服务:

资源简介:

This dataset contains LLM-based classifications of public IT procurement awards extracted from Tenders Electronic Daily (TED), the EU’s official public procurement portal. The dataset focuses on IT-related procurement awards in CPV divisions 48 (software packages and information systems) and 72 (IT services, consulting, software development, internet and support). Each record represents one procurement award with identifiable buyer and supplier information, normalized contract value, TED title and description fields, and an LLM-as-a-judge classification into three theoretically grounded types of digital change: Digitization, Digitalization, and Digital Transformation. The classification was generated using a zero-shot, rubric-based LLM prompt. For each award, the model received the procurement title and description and returned three scores ranging from 0 to 100, one for each digital-change dimension, plus a short textual reasoning. The final class label corresponds to the highest-scoring dimension. The dataset is intended for research on digital government, public procurement, GovTech markets, public-sector innovation, and AI-assisted analysis of administrative data. The dataset should be interpreted as a classification of procurement descriptions and procurement intent, not as evidence of actual implementation outcomes or realized digital transformation. Full tender documentation, project execution data, and ex post performance information are not included.

本数据集包含基于大语言模型(LLM)的公开IT采购中标项目分类结果,数据源自欧盟官方公共采购门户Tenders Electronic Daily(TED)。本数据集聚焦于CPV分类第48大类(软件包与信息系统)及第72大类(IT服务、咨询、软件开发、互联网及支持服务)相关的IT采购中标项目。每条记录对应一则采购中标信息,包含可识别的采购方与供应商信息、标准化合同金额、TED的标题与描述字段,以及由大语言模型作为评判者生成的三类基于理论框架的数字化变革分类:数字化(Digitization)、数字化转型(Digitalization)、数字转型(Digital Transformation)。 该分类通过基于评分准则的零样本(zero-shot)大语言模型提示生成。针对每一则中标项目,模型将接收采购标题与描述文本,并返回三个介于0至100的分值(对应每一类数字化变革维度),以及一段简短的文本推理依据。最终分类标签对应得分最高的维度。本数据集旨在服务于数字政务、公共采购、GovTech市场、公共部门创新以及行政数据的AI辅助分析等领域的研究。 本数据集仅可用于对采购描述与采购意图进行分类,不应被视为实际实施成果或已实现数字转型的佐证。完整招标文档、项目执行数据以及事后绩效信息均未纳入本数据集。

提供机构:
Zenodo
创建时间:
2026-05-14
二维码
社区交流群
二维码
科研交流群
商业服务