Visualization-Referenced Dataset

该数据集旨在从现有的社区问答（CQA）数据集中筛选出具有代表性的数据，重点关注合适的图表分布和细粒度属性，以便有效训练多语言大模型（MLLMs）。该数据集解决了现有CQA数据中的分布问题，并采用了一种数据引擎进行筛选和生成，以优化数据分布。其规模适中，涵盖了图表特征和问答任务的真实世界范围。该数据集的任务是图表问答（Cqa）。

This dataset aims to select representative data from existing Community Question Answering (CQA) datasets, with a focus on appropriate chart distribution and fine-grained attributes, so as to effectively train Multilingual Large Language Models (MLLMs). It addresses the distribution issues in current CQA datasets, and employs a data engine for data screening and generation to optimize the data distribution. Boasting a moderate scale, it covers the real-world scope of chart features and question answering tasks. The task of this dataset is Chart Question Answering (CQA).