qcri-ai/HCTQA

Human Centric Tables Question Answering (HCTQA)是一个用于评估大型语言模型在处理复杂、以人为本的表格上的问答性能的基准数据集。该数据集包括现实世界和合成的表格，以及与之相关的问题和逻辑结构元数据。数据集中的问题具有不同的复杂度，需要模型处理复杂的结构推理、数值聚合和依赖上下文的理解。

Human Centric Tables Question Answering (HCTQA) is a benchmark dataset designed to evaluate the performance of large language models on question answering over complex, human-centric tables. The dataset consists of both real-world and synthetic tables, along with associated questions and logical structural metadata. The questions in the dataset vary in complexity, requiring models to handle complex structure reasoning, numeric aggregation, and context-dependent understanding.