[CFP] DMAIL 1st International Workshop on Data Mining and AI for Law

22 views
Skip to first unread message

Sabine Wehnert

unread,
Jun 9, 2025, 3:49:00 PMJun 9
to Jurix Foundation for Legal Knowledge Based Systems

Call for Papers: DMAIL 2025 – The First International Workshop on Data Mining and AI for Law @ICDM Conference 2025

We are pleased to announce the First International Workshop on Data Mining and AI for Law (DMAIL 2025), which will take place on November 12, 2025, in Washington, DC, USA. This workshop aims to bring together researchers, practitioners, and policymakers to explore innovative applications of data mining and artificial intelligence in the legal domain.

Topics of interest include but are not limited to: legal decision prediction, legal text summarization, bias and fairness in legal AI, argument mining, case law retrieval, and AI-assisted legal reasoning.

Important Dates:

  • Paper Submission Deadline: September 1, 2025

  • Notification of Acceptance: September 15, 2025

  • Camera-ready Papers Due: September 25, 2025

  • Workshop Date: November 12, 2025

Submissions should be no longer than 10 pages using the IEEE 2-column format and will undergo a double-blind peer review process. Authors are encouraged to provide supplementary code and data for reproducibility.

For full details and paper submission, see the attached poster and visit: https://dmail-workshop.github.io/DMAIL2025/

Submit your paper via EasyChair: https://easychair.org/conferences?conf=dmail2025

DMAIL_2025_cfp.pdf

Sabine Wehnert

unread,
Aug 20, 2025, 12:23:43 AMAug 20
to Jurix Foundation for Legal Knowledge Based Systems

Dear colleagues,

Following our initial announcement, we are excited to share a major new component of DMAIL 2025: The First International Workshop on Data Mining and AI for Law, a brand new legal challenge!

We are pleased to announce and invite participation in the Legal Bias Flagging Challenge, a competition designed to push the boundaries of AI in flagging bias within legal texts. This challenge is held in conjunction with our main workshop, which will take place at the 25th IEEE International Conference on Data Mining (ICDM 2025) in Washington DC, USA (November 12, 2025).

The DMAIL 2025 Legal Bias Flagging Challenge

This competition is an opportunity for researchers and practitioners to test their models on a novel and ethically critical task.

The Task: Participants are challenged to build a model that can automatically flag potential legal bias. For a legal case, participants will be shown a specific text passage and must provide a binary output (yes/no) indicating the presence of bias within that passage, as well as an automatically generated reasoning for the output label. The full case text will also be provided for broader context.

The Data: To simulate a challenging scenario, we will provide a dataset of legal cases from multiple jurisdictions, including the USA, Germany, Japan, Vietnam, and Korea. We will provide this data without any training instances. Participants are expected to develop their models based on their own methodologies and general knowledge.

The Evaluation: The DMAIL Organizing Group possesses a hidden ground truth for all instances in the dataset. Submissions will be evaluated against this ground truth to determine the performance of each model and determine a winner.

Challenge Rules and Evaluation

The challenge has been designed to prioritize the flagging of bias, ensuring that models are sensitive to its presence.

Evaluation Metric: Submissions will be primarily evaluated using an F2-score, which prioritizes recall over precision for the positive label (yes). This metric is well-suited for the task of flagging potential bias within pre-selected text passages.

Winning Categories: To recognize a diverse range of approaches and encourage accessible solutions, we will announce winners in two distinct categories:

  • Overall Winner: Awarded to the participant with the best overall score on the challenge dataset.

  • Low-Resource Setting Winner: Awarded to the participant who achieves the highest score while adhering to specific low-resource constraints.

Low-Resource Setting Criteria: To qualify for this category, participants must meet the following criteria and provide a clear description of their setup:

  • Model Availability: Any pre-trained employed model must be publicly available (e.g., on HuggingFace, Ollama, or similar platforms).

  • Model Size and VRAM: The model should meet at least one of the following constraints:

    • The model size must not exceed a specific size threshold (100M parameters).

    • The model must not exceed a maximum VRAM usage (16GB).

  • Transparency: Participants must clearly describe their experimental setup, including the hardware used, VRAM requirements, prompts, and any other relevant technical details in their paper.

Why Participate? This challenge provides a unique platform to contribute to the growing field of responsible AI for law. Participants will gain experience working with a diverse, multilingual legal dataset and contribute to a critical discussion on fairness and transparency in legal systems. The winners will be recognized at the DMAIL 2025 workshop.

Challenge Key Dates:

  • Challenge Registration Opens: August 10, 2025

  • Challenge Dataset Release: August 12, 2025

  • Submission Deadline: August 30, 2025

  • Results & Final Ranking Announcement: September 1, 2025

Challenge Paper Submission: We highly encourage participants to submit a paper describing their methodology and results from the challenge. A separate review cycle will be held for these papers.

  • Challenge Paper Submission Deadline: September 5, 2025

  • Notification to Challenge Authors: September 18, 2025

  • Challenge Paper Camera-ready Deadline: September 25, 2025

For those interested in submitting a research paper to the main workshop, please note our upcoming deadlines:

    • Paper Submission Deadline: September 1, 2025

    • Notification to Authors: September 15, 2025

    • Camera-ready and Copyright Deadline: September 25, 2025

    • Workshop Date: November 12, 2025

      We would be grateful if you could circulate this announcement among your networks. Detailed challenge guidelines and submission instructions for both the workshop and the competition are available on our website: https://dmail-workshop.github.io/DMAIL2025/

      Thank you for your support, and we look forward to your participation in DMAIL 2025 and our inaugural challenge!

      Kind regards, Haihua Chen, Ken Satoh, Ha Thanh Nguyen, Joel Niklaus, Yang Zhang, and Sabine Wehnert DMAIL 2025 Chairs - https://dmail-workshop.github.io/DMAIL2025/organization/

      Reply all
      Reply to author
      Forward
      0 new messages