Skip to content
Home » First Accounting Co., Ltd. First Accounting’s technical paper on document image analysis was accepted at the international conference LREC-COLING 2024

First Accounting Co., Ltd. First Accounting’s technical paper on document image analysis was accepted at the international conference LREC-COLING 2024

[First Accounting Co., Ltd.] A technical paper on First Accounting’s document image analysis was presented at the international conference LREC-COLING.
Adopted in 2024

*View in browser* *First Accounting Co., Ltd. *
Press release: March 27, 2024
**
A technical paper on document image analysis for first accounting was accepted at the international conference LREC-COLING 2024
*Proposes a technology that more flexibly analyzes imaged documents by using large-scale language models*
* First Accounting Co., Ltd. (Head office: Minato-ku, Tokyo, President and CEO: Mori
Keitaro (hereinafter referred to as First Accounting), co-founder and chief executive officer of our company, will be speaking at the Joint International Conference on Computational Linguistics, Language Resources and Evaluation to be held in Turin, Italy from May 20th to 25th, 2024. FA led by Research Scientist (CRS) Masato Fujitake We are pleased to announce that the Research paper has passed peer review and been accepted. *
* ■Summary of the paper*
The recently accepted paper proposes a more flexible method for analyzing imaged documents using large-scale language models (LLM). Up until now, document understanding has been strengthened by incorporating pre-learning on images, text, and layout structure, but this requires fine-tuning for each task or dataset, resulting in high costs for model training and operation. It was a burden.
We therefore proposed a new model that integrates the strengths of existing research in document image understanding with the superior language understanding ability of LLM and is fine-tuned on a multimodal instruction dataset. This is a single model that performs document image understanding.
Through experiments, improvements were made over the baseline model in various document analysis tasks, such as extracting information on items to be read from document images, classifying document image types, and document image visual question answering that flexibly answers questions about document images. It has been proven that it does. Additionally, it was confirmed that not only the performance of document image tasks improved, but also the performance of
general-purpose natural language processing tasks.
* ■About LREC-COLING 2024*
It is a joint international conference on computational linguistics, language resources and evaluation, jointly sponsored by the ELRA Language Resources Association (ELRA) and the International Commission on Computational Linguistics (ICCL), two major international organizations in the field of computational linguistics. It will be held in Turin, Italy from May 20th to 25th, 2024. In addition to the three-day plenary session, there will also be a total of three days of workshops and tutorials before and after the conference.
This hybrid international conference will bring together researchers and practitioners in computational linguistics, speech, multimodality, and natural language processing, with a particular focus on evaluation and the development of resources to support work in these areas. Continuing the tradition of the well-established parent conferences COLING and LREC, this joint conference will feature a grand challenge and provide ample opportunity for participants to exchange information and ideas through oral presentations and an extensive poster session. https://lrec-coling-2024.org/about-lrec-coling/
* ■First Accounting Comments*
We are pleased that our research on large-scale language models has progressed and that our research has been accepted by a peer-reviewed international academic conference. In this research, we focused on English document images, but we would like to further develop the technology cultivated in this research and expand it to include multilingual support.
(Co-founder/Chief Research Scientist (CRS) Masato Fujitake)
* ■About First Accounting*
First Accounting is a company that provides services for corporations to utilize the power of AI to streamline and automate accounting operations.
In character recognition technology, he researches not only the computer vision field, but also generative AI such as LLM and the latest technologies, and presents papers at various academic conferences. We develop services based on these research results and provide them to many large corporate customers and accounting vendors. Our purpose is to “instill confidence and courage by removing constraints.” We will use AI technology to remove various constraints on our customers’ operations and do our best to help them focus on more valuable operations.

Company name: First Accounting Co., Ltd. (TSE Growth: 5588)
Address: 3rd floor, VORT Hamamatsucho I, 1-6-15 Hamamatsucho, Minato-ku, Tokyo Established: June 2016
Representative: Keitaro Mori, Representative Director and President URL: https://www.fastaccounting.jp/

* ■Inquiries regarding this matter*
First Accounting Co., Ltd.
Person in charge: Public Relations
E-mail: press@fastaccounting.co.jp
All product and company names are trademarks or registered trademarks of their respective owners.

*About details about this release*
https://prtimes.jp/main/html/rd/p/000000107.000061842.html