[CFP] SemEval 2025 Task 5 - LLMs4Subjects - 1st Call for Shared Task Participation

9 views
Skip to first unread message

Jennifer D'Souza

unread,
Oct 6, 2024, 11:24:34 AMOct 6
to ISSI listserv
[apologies if you receive multiple copies of this call]

Dear colleagues and friends,

We are pleased to release the 1st Call for Participation to the LLMs4Subjects Shared Task organized as part of SemEval 2025.

Overview: As the first of its kind, LLMs4Subjects invites the research community to develop cutting-edge LLM-based semantic solutions for the subject tagging of the Leibniz University's Technical Library's open-access collection. The shared task provides an opportunity for the research community to creatively utilize LLMs for subject tagging of technical records. Systems need to demonstrate bilingual language modeling in understanding technical documents in both German and English. Moreover, successful solutions may be directly integrated into the operational workflows of the TIB Leibniz Information Centre for Science and Technology University Library.

What we provide to participants: a human-readable form of a subject's taxonomy (this is the GND or Gemeinsame Normdatei, the integrated authority file used for cataloging in German-speaking countries) and a large collection of technical records tagged with these subjects from the TIB's open-access collection called TIBKAT.

More details on the task website: https://sites.google.com/view/llms4subjects/

LLMs4Subjects defines the following three tasks:

- Task 1: Learn the GND
- Task 2: Align subject tagging to the TIBKAT collection
- (Optional and Fun) Task 3: Develop Elegant Frontend Interfaces for Subject Tagging

The task offers a broad range of research questions that can be explored. View them here. We encourage interested teams to sign up and begin development as soon as possible. Systems can be developed over a 3-month period, with the official evaluation phase starting on January 10, 2025.

LLMs4Subjects will have three separate evaluations:

- Evaluation 1: Quantitative Metrics-based Evaluations
- Evaluation 2: Qualitative Evaluations by the Human Subject Specialists
- (Optional Task) Evaluation 3: HCI evaluations for subject indexing interfaces submitted

To participate in the LLMs4Subjects shared task,
 
1. please submit your interest to participate using our online form (https://forms.gle/YQzupcoySAyJi45c6), 
2. sign up to the shared task Google Groups (https://groups.google.com/u/6/g/llms4subjects) for FAQs, news, and announcements, and,
3. last but not the least, download the datasets (https://github.com/jd-coderepos/llms4subjects/) to begin development.

Dates

Training and validation datasets available: October 2, 2024
Test data available/Evaluation starts: January 10, 2025
Evaluation ends: January 31, 2025
Participant paper submissions due: February 28, 2025
Notification to authors: March 31, 2025
Camera ready due: April 21, 2025
SemEval workshop: TBD

Task Organizers

Jennifer D'Souza, Sameer Sadruddin, Holger Israel, Mathias Begoin et al.
All organizers are affiliated with the TIB Leibniz Information Centre for Science and Technology - Germany (https://www.tib.eu/en/)

We look forward to having you on board!

Contact: llms4subjects [at] gmail.com
Reply all
Reply to author
Forward
0 new messages