DeepWTO

DeepWTO is the first AI open source project for the legal professionals.

Our goal is to build public datasets with the world trade law system and stimulates the global research collaboration for application of AI technology starting from the world trade law system to any other legal systems in the world.

Although there has been a significant advancement in the field of Natural Language Processing (NLP) with Deep Learning (DL) for past few years, however, any commercial services or products for legal profession has not been released or proven its usefulness to the public leveraging those technologies. This scarcity of implementation of AI technologies to the legal systems is generally due to the lack of available public datasets.

Law itself is generally local and requiring high level of expert knowledge. Thus it was relatively hard for the research community to build public datasets that the community can academically research together. Therefore, we decided to build public datasets in the world trade law system and prove its applicability of AI technologies to those datasets.

World trade law system, currently maintained and enforced by the World Trade Organization (WTO), is very symbolic international regime since the global community has maintained it with a high priority as a societal proof of global peace & collaboration after World War II.

Our project will develop towards the two main directions:

First, prepares public datasets in world trade law system that are fit to various applications of different legal AI-NLP tasks - such as predicting invokable articles for given factual aspect, predicting legality of given factual aspect in terms of invoked article, calculation of document similarity to assist case search etc.

Second, find the most effective Al model architecture that could be applied for the given task with the prepared dataset from the first direction.

For the first direction, the project already had prepared its public dataset for the task of predicting invokable article with given factual aspect in world trade law system. Check link below:
[dataset-api] https://github.com/deepwto/deepwto-api
[direct download] https://drive.google.com/drive/u/2/folders/1BpwYLqSBXxSgv8cmItwbohIkfebJr3lX

For the second direction, we also had performed a research and find an effective AI model architecture that could predict invokable legal article for given factual aspect. After train the AI model, we had hosted the model as an interactive demo where any legal practitioners can check how model predicts for each different cases of WTO.  Check below links:
[demo] https://www.deepwto.org
[paper] https://drive.google.com/file/d/16zXC9GBdoXqgcdLn7jeTenDJVdPuL72U/view
[code] https://github.com/DeepWTO/deepwto-draft

The project page for developers is as below. Every code is updated as a public open source.
https://github.com/DeepWTO

We currently plans to submit above two achievements - 1) the preparation of the public dataset and 2) finding of AI model structure that effectively works for predicting invokable articles for given factual aspect - to 2021 ICAIL (International Conference on Artificial Intelligence and Law). ICAIL is the most authoritative conference with the combined two topics, AI and Law. For the submission purpose and upgrade of the AI model, we are currently raising the funds. Please support our project so that we can step forward and share our research work with the global community.

For more information about the ICAIL, please check the below:
[IAAIL - the association that holds the ICAIL] http://www.iaail.org/
[2021 ICAIL Venue Announcement] http://www.iaail.org/?q=article/iaail-executive-committee-announces-venue-icail-2021

We believe the application of AI technology in legal systems will help legal professionals to work more effectively so that more people in our society can benefit from the legal system.