Ruqii/t1c-review-mining

该数据集是T1-C基准测试理念的轻量级种子数据集，旨在从电子商务评论中挖掘结构化见解。数据来源于McAuley-Lab/Amazon-Reviews-2023，包含20,000条评论行的样本切片和完整的产品元数据表。选择这个数据切片的原因是它大小适中、易于复制，并且有足够的测试量来评估情感分析、投诉提取、方面提取、聚类和产品级聚合等任务。数据集还提供了评论行的详细字段，并建议了下一步的研究方向，如构建人工标注的金标准集和定义具体任务。

This folder contains a lightweight seed dataset for the T1-C benchmark idea: mining ecommerce reviews into structured insights. The source dataset is McAuley-Lab/Amazon-Reviews-2023, and it includes a sample slice of 20,000 review rows and the full product metadata table. This slice was chosen because it is small enough to work with quickly, public and easy to reproduce, and has enough volume to test sentiment, complaint extraction, aspect extraction, clustering, and product-level aggregation. The dataset also provides detailed fields for review rows and suggests next steps such as building a gold set of human-labeled reviews and defining specific tasks.