意大利用户请求和对应的约束表示试点数据集
收藏资源简介:
该数据集是专门为研究可再生能源社区(REC)中的用户偏好转换成能量优化约束而创建的。它包含了26个意大利用户的特定用能偏好的请求,这些请求被标注为对应的约束表示。数据集来源于更大规模的用户请求和标注意图语料库,用于训练对话代理的自然语言理解模块。每个请求都经过两步手动标注,第一步是识别和标记表达偏好的文本跨度,第二步是将这些偏好转换为可能的约束表示。
This dataset is specifically developed for research on converting user preferences into energy optimization constraints within Renewable Energy Communities (REC). It includes 26 requests from Italian users detailing their specific energy consumption preferences, with each request annotated with corresponding constraint representations. The dataset is derived from a larger-scale corpus of user requests with annotated intents, which was utilized to train the natural language understanding (NLU) module of conversational agents. Each request undergoes two-stage manual annotation: the first stage involves identifying and tagging the text spans that express user preferences, while the second stage converts these preferences into potential constraint representations.

- 1From User Preferences to Optimization Constraints Using Large Language Models卡利亚里大学,数学与计算机科学系,卡利亚里,意大利 · 2025年



