Spaces:
Running
Running
extend readme for organizational card
Browse files
README.md
CHANGED
|
@@ -7,23 +7,40 @@ sdk: static
|
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
-
# Open Multi-Label ASJC Classification
|
| 11 |
|
| 12 |
This Team Space hosts the **open, multi-label implementation of the All Science Journal Classification (ASJC) taxonomy**, designed to classify scientific documents at the individual level.
|
| 13 |
|
| 14 |
-
##
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
Traditional ASJC classification is limited by incomplete sources, journal-level labels, and single-label assignments. This project provides:
|
| 16 |
- **Multi-label classification across 307 subjects**
|
| 17 |
- Fine-tuned **SciBERT model** trained on Crossref metadata
|
| 18 |
- Methods for **collection-level analysis** (researcher portfolios, institutions, datasets)
|
| 19 |
|
| 20 |
-
## Features
|
| 21 |
-
- High
|
| 22 |
- Works with or without source title metadata
|
| 23 |
- Open, reproducible, and ready for research use
|
| 24 |
|
| 25 |
-
##
|
| 26 |
-
-
|
| 27 |
- Example datasets and label-averaging utilities
|
| 28 |
|
| 29 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7 |
pinned: false
|
| 8 |
---
|
| 9 |
|
| 10 |
+
# 🧠 Open Multi-Label ASJC Classification
|
| 11 |
|
| 12 |
This Team Space hosts the **open, multi-label implementation of the All Science Journal Classification (ASJC) taxonomy**, designed to classify scientific documents at the individual level.
|
| 13 |
|
| 14 |
+
## 👥 Team
|
| 15 |
+
- **Michael Gusenbauer** – Johannes Kepler University Linz | ORCID: [https://orcid.org/0000-0001-7768-2351](https://orcid.org/0000-0001-7768-2351)
|
| 16 |
+
- **Jochen Endermann** – University of Applied Sciences Kufstein
|
| 17 |
+
- **Harald Huber** – University of Applied Sciences Kufstein
|
| 18 |
+
- **Simon Strasser** – University of Applied Sciences Kufstein
|
| 19 |
+
- **Andreas-Nizar Granitzer** – Norwegian Geotechnical Institute | ORCID: [https://orcid.org/0000-0002-5839-4300](https://orcid.org/0000-0002-5839-4300)
|
| 20 |
+
- **Thomas Ströhle** – Universität Innsbruck | ORCID: [https://orcid.org/0000-0002-1954-6412](https://orcid.org/0000-0002-1954-6412)
|
| 21 |
+
|
| 22 |
+
## 🎯 Purpose
|
| 23 |
Traditional ASJC classification is limited by incomplete sources, journal-level labels, and single-label assignments. This project provides:
|
| 24 |
- **Multi-label classification across 307 subjects**
|
| 25 |
- Fine-tuned **SciBERT model** trained on Crossref metadata
|
| 26 |
- Methods for **collection-level analysis** (researcher portfolios, institutions, datasets)
|
| 27 |
|
| 28 |
+
## ✨ Features
|
| 29 |
+
- High performance: weighted F1-score 0.892 (307 subjects) | 0.934 (26 parent subjects)
|
| 30 |
- Works with or without source title metadata
|
| 31 |
- Open, reproducible, and ready for research use
|
| 32 |
|
| 33 |
+
## 🗂 Content
|
| 34 |
+
- Model and code for reproducing results
|
| 35 |
- Example datasets and label-averaging utilities
|
| 36 |
|
| 37 |
+
## 📖 Citation
|
| 38 |
+
If you use this work, please cite:
|
| 39 |
+
|
| 40 |
+
```bibtex
|
| 41 |
+
@article{gusenbauer2025open,
|
| 42 |
+
author = {Gusenbauer, Michael and Endermann, Jochen and Huber, Harald and Strasser, Simon and Granitzer, Andreas-Nizar and Ströhle, Thomas},
|
| 43 |
+
title = {Open Multi-Label Implementation of the All Science Journal Classification (ASJC) Taxonomy},
|
| 44 |
+
journal = {Scientometrics},
|
| 45 |
+
year = {2025}
|
| 46 |
+
}
|