CGIAR/Chi-Metrics-2024

这是一个包含2,300条来自肯尼亚Farmer.Chat平台的匿名用户互动记录的数据集，该平台是一个AI驱动的农业咨询服务平台，数据收集时间为2023年10月至2024年4月。数据涵盖了查询类型、响应质量、用户参与模式以及系统性能等多个农业价值链（乳制品、咖啡、马铃薯等）的指标。数据集包括响应忠实度、相关性评分、查询清晰度指标、认知需求水平等定量测量，以及按用户类型（农业扩展代理、带头农户、农民和农业企业家）分类的匿名用户人口统计信息。数据集通过Flesch-Kincaid可读性评分评估查询复杂性，通过多个指标（包括上下文精确度和RAGAS评分）评估响应质量。该数据集为研究小农户和农业顾问如何与AI驱动的农业咨询系统互动提供了有价值的见解，尤其是在资源匮乏的环境中，并可以指导开发更加便捷有效的农业AI应用程序。

This dataset consists of 2,300 anonymized user interaction records from the Farmer.Chat platform, an AI-powered agricultural advisory service deployed in Kenya between October 2023 and April 2024. The data encompasses metrics across various agricultural value chains such as dairy, coffee, potato, and other crops, including query types, response quality, user engagement patterns, and system performance. The dataset includes quantitative measures of response fidelity, relevance scores, query clarity metrics, cognitive demand levels, and anonymized demographic information of users categorized by types (Agriculture Extension Agents, Lead Farmers, Farmers, and Agripreneurs). Query complexity is evaluated using the Flesch-Kincaid readability score, and response quality is assessed through multiple metrics including context precision and RAGAS scores. The dataset offers valuable insights into how smallholder farmers and agricultural advisors interact with AI-powered agricultural advisory systems, especially in low-resource settings, and can inform the development of more accessible and effective agricultural AI applications.