LLMs4Subjects - Subject Tagging for a Technical Library
Subject indexing of technical bibliographic records for the TIB Leibniz Information Centre's open-access TIBKAT catalog. Annotators assign subject headings and record types to English/German technical records, mirroring the library workflow SemEval-2025 Task 5 automates with LLMs. This showcase is a simplified adaptation of that task.
Configuration Fileconfig.yaml
This Potato config reproduces the annotation task. Save it as config.yaml and run potato start config.yaml to try it.
# LLMs4Subjects - Subject Tagging for a Technical Library
# Based on D'Souza et al., SemEval-2025 Task 5
# Paper: https://aclanthology.org/2025.semeval-1.328/
# Dataset: https://github.com/jd-coderepos/llms4subjects
#
# SemEval-2025 Task 5 (LLMs4Subjects) asks systems to recommend subject
# headings from the GND (Gemeinsame Normdatei) vocabulary for technical
# bibliographic records in the TIB Leibniz Information Centre's bilingual
# (English/German) open-access TIBKAT catalog. This showcase is a
# simplified HUMAN-annotation adaptation: rather than the full open GND
# vocabulary of tens of thousands of headings, annotators tag each record
# with broad subject domains and its record type, reproducing the manual
# subject-indexing workflow the shared task automates.
#
# Note: the official gold standard uses the full GND subject taxonomy;
# the fixed domain list below is a demonstration simplification.
annotation_task_name: "LLMs4Subjects - Subject Tagging for a Technical Library"
task_dir: "."
data_files:
- sample-data.json
item_properties:
id_key: "id"
text_key: "text"
output_annotation_dir: "annotation_output/"
output_annotation_format: "json"
port: 8000
server_name: localhost
annotation_schemes:
- annotation_type: multiselect
name: subject_categories
description: "Select all subject domains this record should be indexed under (a simplified stand-in for GND subject headings)."
labels:
- "Computer Science"
- "Mathematics"
- "Physics"
- "Biology"
- "Medicine"
- "Engineering"
- "Social Science"
- "Humanities"
- "Law"
- "Economics"
tooltips:
"Computer Science": "Algorithms, programming, AI, databases, software, etc."
"Mathematics": "Pure or applied mathematics, statistics, logic"
"Physics": "Classical, quantum, astrophysics, particle physics, etc."
"Biology": "Molecular biology, ecology, genetics, evolution, etc."
"Medicine": "Clinical medicine, pharmacology, public health, etc."
"Engineering": "Mechanical, electrical, civil, chemical engineering, etc."
"Social Science": "Psychology, sociology, political science, anthropology, etc."
"Humanities": "History, philosophy, literature, linguistics, etc."
"Law": "Legal theory, constitutional law, international law, etc."
"Economics": "Micro/macroeconomics, finance, econometrics, etc."
- annotation_type: radio
name: record_type
description: "What type of bibliographic record is this?"
labels:
- "Article"
- "Book"
- "Conference"
- "Report"
- "Thesis"
keyboard_shortcuts:
"Article": "1"
"Book": "2"
"Conference": "3"
"Report": "4"
"Thesis": "5"
tooltips:
"Article": "Journal or periodical article"
"Book": "Monograph or edited book"
"Conference": "Conference or workshop proceedings paper"
"Report": "Technical or research report"
"Thesis": "Dissertation or thesis"
annotation_instructions: |
You are performing subject indexing for a technical-library catalog record
(TIBKAT), in the style of SemEval-2025 Task 5 (LLMs4Subjects). Records may be
in English or German. Your tasks are:
1. Read the record's title and text and identify what it is about.
2. Select all subject domains under which the record should be indexed.
3. Choose the record type (article, book, conference, report, or thesis).
html_layout: |
<div style="padding: 15px; max-width: 800px; margin: auto;">
<div style="background: #fefce8; border: 1px solid #fde68a; border-radius: 8px; padding: 16px; margin-bottom: 16px;">
<strong style="color: #a16207;">Title:</strong>
<p style="font-size: 17px; font-weight: 600; line-height: 1.5; margin: 8px 0 0 0;">{{title}}</p>
</div>
<div style="background: #f0f9ff; border: 1px solid #bae6fd; border-radius: 8px; padding: 16px; margin-bottom: 16px;">
<strong style="color: #0369a1;">Text:</strong>
<p style="font-size: 16px; line-height: 1.7; margin: 8px 0 0 0;">{{text}}</p>
</div>
</div>
allow_all_users: true
instances_per_annotator: 50
annotation_per_instance: 2
allow_skip: true
skip_reason_required: false
Sample Datasample-data.json
[
{
"id": "subj_001",
"text": "We propose a novel transformer architecture for protein folding prediction that achieves state-of-the-art results on the CASP14 benchmark. Our model combines attention mechanisms with geometric deep learning to capture spatial relationships between amino acid residues.",
"title": "Deep Learning Approaches to Protein Structure Prediction"
},
{
"id": "subj_002",
"text": "This paper examines the impact of monetary policy on income inequality across OECD countries from 2000 to 2020. Using panel data regression with fixed effects, we find that expansionary monetary policy disproportionately benefits asset holders.",
"title": "Monetary Policy and Income Inequality: A Cross-Country Analysis"
}
]
// ... and 8 more itemsGet This Design
Clone or download from the repository
Quick start:
git clone https://github.com/davidjurgens/potato-showcase.git cd potato-showcase/semeval/2025/task05-llms4subjects potato start config.yaml
Dataset & paper
D'Souza, Sadruddin, Israel, Begoin, and Slawig. SemEval-2025 Task 5: LLMs4Subjects. SemEval-2025.
Citation (BibTeX)
@inproceedings{dsouza-etal-2025-semeval,
title = "{S}em{E}val-2025 Task 5: {LLM}s4{S}ubjects - {LLM}-based Automated Subject Tagging for a National Technical Library{'}s Open-Access Catalog",
author = "D{'}Souza, Jennifer and Sadruddin, Sameer and Israel, Holger and Begoin, Mathias and Slawig, Diana",
editor = "Rosenthal, Sara and Ros{\'a}, Aiala and Ghosh, Debanjan and Zampieri, Marcos",
booktitle = "Proceedings of the 19th International Workshop on Semantic Evaluation (SemEval-2025)",
month = jul,
year = "2025",
address = "Vienna, Austria",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2025.semeval-1.328/",
pages = "2570--2583",
ISBN = "979-8-89176-273-2"
}Details
Annotation Types
Domain
Use Cases
Tags
Found an issue or want to improve this design?
Open an IssueRelated Designs
Food Hazard Detection
Food safety hazard detection task requiring annotators to classify short food-incident report texts by hazard-category and product-category (coarse labels), and to identify the specific hazard and product mentions (fine-grained labels). Based on SemEval-2025 Task 9: The Food Hazard Detection Challenge.
iSarcasmEval: Intended Sarcasm Detection
SemEval-2022 Task 6 (iSarcasmEval): detecting intended sarcasm in English and Arabic. Unlike prior datasets labeled by third parties, iSarcasmEval texts are labeled by their own authors for intended sarcasm. Annotators decide whether a text is sarcastic and, if so, which category of ironic speech it expresses.
Memotion Analysis - Sentiment and Type Classification of Memes
Classify the overall sentiment of internet memes and identify their communicative types (sarcastic, humorous, offensive, motivational), based on SemEval-2020 Task 8 (Sharma et al.). Annotators analyze both text and image descriptions of memes.