Add paper link and task category

This PR improves the dataset card by:
- Adding the `visual-document-retrieval` task category to the YAML metadata.
- Including a link to the original research paper on the Hugging Face Hub.
- Adding a link to the official GitHub repository.
- Including the BibTeX citation for the dataset.

README.md

@@ -1,74 +1,79 @@
----
-language:
-- en
-tags:
-- diagrams
-- scientific papers
-- retrieval augemented generation
-- multimodal retrieval
-- ICLR
-- ICML
-- NeurIPS
-- TMLR
-pretty_name: DiagramBank
-size_categories:
-- 10K<n<100K
-configs:
-- config_name: default
-  data_files:
-  - split: train
-    path: data.jsonl
-dataset_info:
-  features:
-  - name: platform
-    dtype: string
-  - name: venue
-    dtype: string
-  - name: year
-    dtype: int32
-  - name: title
-    dtype: string
-  - name: abstract
-    dtype: string
-  - name: keywords
-    dtype: string
-  - name: areas
-    dtype: string
-  - name: tldr
-    dtype: string
-  - name: scores
-    sequence: int32
-  - name: decision
-    dtype: string
-  - name: authors
-    dtype: string
-  - name: author_ids
-    dtype: string
-  - name: cdate
-    dtype: string
-  - name: url
-    dtype: string
-  - name: platform_id
-    dtype: string
-  - name: bibtex
-    dtype: string
-  - name: figure_path
-    dtype: string
-  - name: figure_number
-    dtype: string
-  - name: figure_caption
-    dtype: string
-  - name: figure_context
-    dtype: string
-  - name: figure_type
-    dtype: string
-  - name: confidence
-    dtype: float32
----
+---
+language:
+- en
+size_categories:
+- 10K<n<100K
+pretty_name: DiagramBank
+task_categories:
+- visual-document-retrieval
+tags:
+- diagrams
+- scientific papers
+- retrieval augmented generation
+- multimodal retrieval
+- ICLR
+- ICML
+- NeurIPS
+- TMLR
+configs:
+- config_name: default
+  data_files:
+  - split: train
+    path: data.jsonl
+dataset_info:
+  features:
+  - name: platform
+    dtype: string
+  - name: venue
+    dtype: string
+  - name: year
+    dtype: int32
+  - name: title
+    dtype: string
+  - name: abstract
+    dtype: string
+  - name: keywords
+    dtype: string
+  - name: areas
+    dtype: string
+  - name: tldr
+    dtype: string
+  - name: scores
+    sequence: int32
+  - name: decision
+    dtype: string
+  - name: authors
+    dtype: string
+  - name: author_ids
+    dtype: string
+  - name: cdate
+    dtype: string
+  - name: url
+    dtype: string
+  - name: platform_id
+    dtype: string
+  - name: bibtex
+    dtype: string
+  - name: figure_path
+    dtype: string
+  - name: figure_number
+    dtype: string
+  - name: figure_caption
+    dtype: string
+  - name: figure_context
+    dtype: string
+  - name: figure_type
+    dtype: string
+  - name: confidence
+    dtype: float32
+---
 
 # DiagramBank
 
-[**DiagramBank**](https://github.com/csml-rpi/DiagramBank) is a large-scale dataset designed for **Retrieval-Augmented Generation (RAG)** on scientific figures. It aggregates papers and their corresponding diagrams from top AI conferences (ICLR, ICML, NeurIPS, TMLR), providing rich metadata including review scores, acceptance decisions, and figure captions.
+[**DiagramBank**](https://github.com/csml-rpi/DiagramBank) is a large-scale dataset designed for **Retrieval-Augmented Generation (RAG)** on scientific figures. It aggregates 89,422 schematic diagrams curated from top-tier AI publications (ICLR, ICML, NeurIPS, TMLR), providing rich metadata including review scores, acceptance decisions, and figure captions.
+
+Paper: [DiagramBank: A Large-scale Dataset of Diagram Design Exemplars with Paper Metadata for Retrieval-Augmented Generation](https://huggingface.co/papers/2604.20857)  
+Code: [https://github.com/csml-rpi/DiagramBank](https://github.com/csml-rpi/DiagramBank)
 
 ## Dataset Structure
 
@@ -118,7 +123,7 @@ print(dataset[0]['figure_caption'])
 You can download the diagrams and automatically reconstruct the folder structure using our setup script.
 ```bash
 # 1. Download the setup script directly from this repo
-wget [https://huggingface.co/datasets/zhangt20/DiagramBank/resolve/main/download_diagrambank.py](https://huggingface.co/datasets/zhangt20/DiagramBank/resolve/main/download_diagrambank.py)
+wget https://huggingface.co/datasets/zhangt20/DiagramBank/resolve/main/download_diagrambank.py
 ```
 ```bash
 # 2. Run it (downloads ~60GB of diagrams from accepted papers)
@@ -126,7 +131,7 @@ wget [https://huggingface.co/datasets/zhangt20/DiagramBank/resolve/main/download
 export FIG_RAG_DIR=<a scratch folder with at least 60 GB of space>
 ```
 
-```
+```bash
 # 1. Default: Download Accepted papers + Core files (DBs/FAISS)
 python download_diagrambank.py
 
@@ -143,4 +148,19 @@ python download_diagrambank.py
 # python download_diagrambank.py --subset all --no-core
 ```
 
-For a more detailed usage, see [https://github.com/csml-rpi/DiagramBank](https://github.com/csml-rpi/DiagramBank)
+For a more detailed usage, see [https://github.com/csml-rpi/DiagramBank](https://github.com/csml-rpi/DiagramBank)
+
+## Citation
+
+```bibtex
+@misc{zhang2026diagrambank,
+  title        = {DiagramBank: A Large-scale Dataset of Diagram Design Exemplars with Paper Metadata for Retrieval-Augmented Generation},
+  author       = {Zhang, Tingwen and Yue, Ling and Xu, Zhen and Pan, Shaowu},
+  year         = {2026},
+  month        = feb,
+  howpublished = {Research Square},
+  note         = {Preprint (Version 1)},
+  doi          = {10.21203/rs.3.rs-8917857/v1},
+  url          = {https://doi.org/10.21203/rs.3.rs-8917857/v1}
+}
+```