Meet FINQA!

A large-scale dataset with 2.8k financial reports for 8k Q&A pairs to study numerical reasoning with structured and unstructured evidence.

Why FINQA?

HIGH-QUALITY

Financial experts from UpWork

LARGE-SCALE

2.7k Financial reports as evidence for 8k experts annotated Q&A pairs for numerical reasoning.

FULL-EXPLAINABILITY

Fully-annotated reasoning programs & supporting facts.

FINANCE-DOMAIN

Reasoning over financial reports.

Explore

We have designed an interface for you to view the data, please click here to explore the dataset and have fun!

Example

In the task, you are given a financial report containing both text and table, the goal is to answer a question requiring numerical reasoning. An example is shown below:

Download (Train/Test Data, Code)

All the code and data are provided in github. The leaderboard for the public test set is hosted here. The leaderboard for the private test set is hosted here

Statistics

The basic statistics, the distribution of supporting fact sources, and the distribution of program steps of FinQA are as follows:

Paper

Please cite our paper as below if you use the FinQA dataset.


        @inproceedings{chen2021finqa,
          title={FinQA: A Dataset of Numerical Reasoning over Financial Data},
          author={Chen, Zhiyu and Chen, Wenhu and Smiley, Charese and Shah, Sameena and Borova, Iana and Langdon, Dylan and Moussa, Reema and Beane, Matt and Huang, Ting-Hao and Routledge, Bryan R and others},
          booktitle={Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing},
          pages={3697--3711},
          year={2021}
        }

Acknowledgement

We sincerely acknowledge Nancy Wang and Peter Zhong for releasing their FinTabNet dataset.