Dear Colleagues,
We are pleased to announce the first Call for Participation for the SIGTYP 2027 Shared Task on Dictionary Digitization https://sigtyp.github.io/st2027-dicts.html.
A big amount of knowledge about the lexicon of most languages of the world (especially indigenous and minority languages) is only accessible in non-machine readable form. Digitizing this knowledge can enable language technology (from online dictionaries to machine and speech translation) which serves these communities. This shared task aims at pushing automatic digitization methods (OCR, dictionary parsing) further and making them more accessible for communities and linguists to use.
Participate as a System Developer: We invite the NLP community to contribute systems for digitization. If you want to participate in the task, please fill out the following Google form: https://docs.google.com/forms/d/e/1FAIpQLSd5Z6I1hdbKfeoIWFWie0ss5c_HyEBDf4n7... Task description see below.
Participate as a Data Contributor: We invite linguists and communities to contribute dictionaries for digitization. We would be happy to include any dictionaries on endangered and indigenous languages. If you would like a specific dictionary to be digitized, please fill out the following Google form: https://docs.google.com/forms/d/e/1FAIpQLSed1ZKdp9bxU5o-EKGfktJ1xp42heSXBAX4... As a data contributor you will help in creating a sample gold standard for your dictionary to test systems on, and vote for your preferred system for digitization.
Important Dates: Training data release: ↣ 1 Dec 2026 Test data release: ↣ 5 Feb 2027 System submissions are due: ↣ 15 Feb 2027 System papers due: ↣ 1 Mar 2027 Workshop Date: between May 9-14 2027 (TBA), co-located with COLING 2027 Further details see: https://sigtyp.github.io/st2027-dicts.html
The Task: The task is about transforming (scanned or native) PDF files of dictionaries into a machine-readable structured dictionary format.
Sub-Task A (OCR): transforming an input PDF file into a plain text one, preserving the layout. Input: a) pdf (scanned or native) of the dictionary pages which should be digitized, b) pdf (scan or native) of the introductory pages of the dictionary Output: plain text, formatted with layout and markup Evaluation Metrics: WER (Word Error Rate), CER (Character Error Rate), Read Order Edit, Markup F1 Further evaluation aspects: estimated cost, token usage, reproducibility and ease of use of the code.
Sub-Task B (Dictionary Parsing): transforming an input PDF or transcribed text file into a structured format. Input: plain text file preserving layout and markup Output: plain text file, with MDF layout Evaluation metrics: Entry F1 (Entry Detection Accuracy), Field Assignment F1 (Assigning tags/labels), ReadOrderEdit Further evaluation aspects: estimated cost, token usage, reproducibility and ease of use of the code.
Combined Task: transforming an input PDF file into a structured dictionary format. Input: a) pdf (scanned or native) of the dictionary pages which should be digitized, b) pdf (scan or native) of the introductory pages of the dictionary Output: plain text file, with MDF layout
You may submit a system to subtask 1 alone, subtask 2 alone, or both subtasks!
Further information can be found on our web page: https://sigtyp.github.io/st2027-dicts.html
Should you have any further questions, please contact Doreen Osmelak mailto:dosmelak@student.unimelb.edu.au.
Regards, Doreen
On behalf of the SIGTYP 2027 Shared Task organizers: Doreen Osmelak, David Setiawan, Luuk Suurmeijer, Charles Kemp, Andreas Shcherbakov, Terry Regier, Nick Thieberger, Ryan Cotterell, Kat Vylomova