This dataset only consists of linearized underlying data table of charts and their corresponding summaries.
Model that use this dataset: https://huggingface.co/saadob12/t5_C2T_big
Created By:
Kanthara, S., Leong, R. T. K., Lin, X., Masry, A., Thakkar, M., Hoque, E., & Joty, S. (2022). Chart-to-Text: A Large-Scale Benchmark for Chart Summarization. arXiv preprint arXiv:2203.06486.
Paper: https://arxiv.org/abs/2203.06486
Orignal github repo: https://github.com/vis-nlp/Chart-to-text
Abstract from the Paper
Charts are commonly used for exploring data and communicating insights. Generating nat- ural language summaries from charts can be very helpful for people in inferring key in- sights that would otherwise require a lot of cognitive and perceptual efforts. We present Chart-to-text, a large-scale benchmark with two datasets and a total of 44,096 charts cover- ing a wide range of topics and chart types. We explain the dataset construction process and analyze the datasets. We also introduce a num- ber of state-of-the-art neural models as base- lines that utilize image captioning and data-to- text generation techniques to tackle two prob- lem variations: one assumes the underlying data table of the chart is available while the other needs to extract data from chart images. Our analysis with automatic and human eval- uation shows that while our best models usu- ally generate fluent summaries and yield rea- sonable BLEU scores, they also suffer from hallucinations and factual errors as well as dif- ficulties in correctly explaining complex pat- terns and trends in charts.
Note
The original paper published two sub-datasets one collected from statista and the other from pew. The dataset upload here is from statista. Images can be downloaded from the github repo mentioned above.
Langugage
The data is in english and the summaries are in english.
Dataset split
train | valid | test |
---|---|---|
24367 | 5222 | 5222 |
Name of Contributor: Saad Obaid ul Islam