FinLayout 2022 : [FinNLP- 2022] Shared Task on layout document analysis in the financial domain, FinLayout
Call For Papers
We would like to invite you to submit to FinLayout, a shared task on document layout analysis in the financial domain, held in conjunction with IJCAI-ECAI-2022, Messe Wien, Vienna, Austria 23th -25th July, 2022 as part of the FinNLP-2022 workshop.
Shared Task URL: https://sites.google.com/nlg.csie.ntu.edu.tw/finnlp-2022/shared-task-finlayout
Workshop URL: https://sites.google.com/nlg.csie.ntu.edu.tw/finnlp-2022/home
Registration Form: https://docs.google.com/forms/d/1EZTTheA4rLomtLOHKU0xQnraCRc1obuk1IGR1TCqwBc/edit?usp=sharing
The 1st edition of FinLayout introduces a shared task on layout document analysis in the financial domain. Visual features allow to give indications about different aspects of the structure of documents. Therefore, most approaches in document layout analysis rely on image-based methods to extract the structure of each page of the document. Deep neural networks that are developed for computer vision have been proven to be an effective method to analyze layout of document images.
In this shared task, we propose to recognize the layout of financial documents represented through 4 labels: Text, Title, Figure and List.
As training set, we propose to use PubLayNet (), a large dataset (~100GB) of document images, of which the layout is annotated with bounding boxes. The dataset contains overs 1 million PDF articles that are publicly available on PubMed Central, a free full-text archive of biomedical and life sciences journal literature at the U.S. National Institutes of Health's National Library of Medicine.
The idea will be to demonstrate that models trained on PubLayNet containing scientific articles can accurately recognize the layout on different type of documents and typically in the financial domain, which proves the effectiveness of transfer learning.
Zhong, Xu and Tang, Jianbin and Yepes, Antonio Jimeno (2019), "PubLayNet: largest dataset ever for document layout analysis". International Conference on Document Analysis and Recognition (ICDAR)
G. Paaß and I. Konya, Machine learning for document structure recognition, in Modeling, Learning, and Processing of Text Technological Data Structures. Springer, 2011, pp. 221–247.
A. M. Namboodiri and A. K. Jain, Document structure and layout analysis, in Digital Document Processing. Springer, 2007, pp. 29–48.
M. El Haj, P. Rayson, S. Young, and M. Walker, Detecting document structure in a very large corpus of UK financial reports. In Proceedings of the Ninth International Conference on Language Resources and Evaluation (LREC-2014). 2014. pp. 1335-1338.
C. Ramakrishnan, A. Patnia, E. Hovy, and G. A. Burns, Layout-aware text extraction from full-text pdf of scientific articles, Source Code for Biology and Medicine, vol. 7, no. 1, p. 7, May 2012. [Online]. Available: https://doi.org/10.1186/1751-0473-7-7
S. Tuarob, P. Mitra, and C. L. Giles, “A hybrid approach to discover semantic hierarchical sections in scholarly documents,” in Proceedings of 13th International Conference on Document Analysis and Recognition (ICDAR), Aug 2015, pp. 1081–1085.
S. Budhiraja and V. Mago, “A supervised learning approach for heading detection,” CoRR, vol. abs/1809.01477, 2018. [Online]. Available: http://arxiv.org/abs/1809.01477
K. Anoop R and R. Christian and G. Cordula and B. Sebastian and B. Steffen and H. Johannes and F. Jean Baptiste, Chargrid: Towards Understanding 2D Documents, In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, 2018. pp.4459--4469.
Xu, Yiheng and Li, Minghao and Cui, Lei and Huang, Shaohan and Wei, Furu and Zhou, Ming, LayoutLM: Pre-training of Text and Layout for Document Image Understanding, In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 2020.
Aggarwal, Milan and Sarkar, Mausoom and Gupta, Hiresh and Krishnamurthy, Balaji, Multi-Modal Association based Grouping for Form Structure Extraction, In Proceedings of 2020 IEEE Winter Conference on Applications of Computer Vision (WACV), 2020, 2064-2073.
Lin W. et al. ViBERTgrid: A Jointly Trained Multi-modal 2D Document Representation for Key Information Extraction from Documents. In: Lladós J., Lopresti D., Uchida S. (eds) Document Analysis and Recognition – ICDAR 2021. Lecture Notes in Computer Science, vol 12821. Springer.
To register your interest in participating in FinLayout shared task, please use the following google form: https://docs.google.com/forms/d/1EZTTheA4rLomtLOHKU0xQnraCRc1obuk1IGR1TCqwBc/edit?usp=sharing
A USD$1000 prize will be rewarded to the best-performing teams.
April 12, 2022: First announcement of the shared task and beginning of registration
April 20, 2022 : Release of training set & scoring scripts.
May 20, 2022: Release of test set.
May 26, 2022: System's outputs submission deadline.
May 30, 2022: Release of results.
May 30, 2022: Shared task title and abstract due
June 06, 2022: Shared task paper submissions due
June 17, 2022: Registration deadline.
June 17, 2022: Camera-ready version of shared task paper due
July 23-25, 2022: FinNLP-2022 workshop @IJCAI-ECAI-2022
For any questions on the shared task, please contact us on firstname.lastname@example.org.
=====Shared Task Co-organizers - Fortia Financial Solutions=====
Abderrahim AIT AZZI