Link dataset to paper and GitHub repository, add Python API usage
#1
by nielsr HF Staff - opened
README.md
CHANGED
|
@@ -1,18 +1,18 @@
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
license: mit
|
|
|
|
|
|
|
| 3 |
task_categories:
|
| 4 |
- text-generation
|
| 5 |
- text-retrieval
|
| 6 |
-
|
| 7 |
-
- en
|
| 8 |
tags:
|
| 9 |
- related-work-generation
|
| 10 |
- scholarly-positioning
|
| 11 |
- citation-evaluation
|
| 12 |
- retrieval-augmented-generation
|
| 13 |
-
pretty_name: RWGBench
|
| 14 |
-
size_categories:
|
| 15 |
-
- 10K<n<100K
|
| 16 |
---
|
| 17 |
|
| 18 |
# RWGBench
|
|
@@ -22,6 +22,9 @@ citation-centric scholarly positioning task. It tests whether a system can
|
|
| 22 |
select, organize, and frame prior work for a target paper, rather than only
|
| 23 |
producing fluent text that resembles a reference related work section.
|
| 24 |
|
|
|
|
|
|
|
|
|
|
| 25 |
## Files
|
| 26 |
|
| 27 |
| File | Entries | Description |
|
|
@@ -101,6 +104,29 @@ data/
|
|
| 101 |
|
| 102 |
Then run the generation and evaluation scripts from the code repository.
|
| 103 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 104 |
## Data Collection And Use
|
| 105 |
|
| 106 |
RWGBench is built from public scholarly documents and metadata. Source
|
|
@@ -108,4 +134,4 @@ documents may have heterogeneous licenses, so users should follow the license
|
|
| 108 |
terms of the underlying papers when redistributing document-derived text.
|
| 109 |
|
| 110 |
The benchmark is intended for research on retrieval-augmented generation,
|
| 111 |
-
citation selection, scholarly writing evaluation, and related work generation.
|
|
|
|
| 1 |
---
|
| 2 |
+
language:
|
| 3 |
+
- en
|
| 4 |
license: mit
|
| 5 |
+
size_categories:
|
| 6 |
+
- 10K<n<100K
|
| 7 |
task_categories:
|
| 8 |
- text-generation
|
| 9 |
- text-retrieval
|
| 10 |
+
pretty_name: RWGBench
|
|
|
|
| 11 |
tags:
|
| 12 |
- related-work-generation
|
| 13 |
- scholarly-positioning
|
| 14 |
- citation-evaluation
|
| 15 |
- retrieval-augmented-generation
|
|
|
|
|
|
|
|
|
|
| 16 |
---
|
| 17 |
|
| 18 |
# RWGBench
|
|
|
|
| 22 |
select, organize, and frame prior work for a target paper, rather than only
|
| 23 |
producing fluent text that resembles a reference related work section.
|
| 24 |
|
| 25 |
+
* **Paper:** [RWGBench: Evaluating Scholarly Positioning in Related Work Generation](https://huggingface.co/papers/2606.24894)
|
| 26 |
+
* **Repository:** [BFTree/RWGBench](https://github.com/BFTree/RWGBench)
|
| 27 |
+
|
| 28 |
## Files
|
| 29 |
|
| 30 |
| File | Entries | Description |
|
|
|
|
| 104 |
|
| 105 |
Then run the generation and evaluation scripts from the code repository.
|
| 106 |
|
| 107 |
+
## Python API Usage
|
| 108 |
+
|
| 109 |
+
To use the dataset with the evaluator script from the code repository:
|
| 110 |
+
|
| 111 |
+
```python
|
| 112 |
+
from src.evaluation.single_paper_evaluator import SinglePaperEvaluator
|
| 113 |
+
|
| 114 |
+
evaluator = SinglePaperEvaluator(
|
| 115 |
+
gold_papers_path="data/gold100_papers.json",
|
| 116 |
+
corpus_path="data/corpus.json",
|
| 117 |
+
use_llm_judge=False,
|
| 118 |
+
citation_frame_model_path="Anonymous2876/rwgbench-citation-frame-classifier",
|
| 119 |
+
)
|
| 120 |
+
|
| 121 |
+
result = evaluator.evaluate(
|
| 122 |
+
paper_id=123,
|
| 123 |
+
generated_text="Prior work has studied retrieval-augmented generation [1].",
|
| 124 |
+
citation_list=[10001], # citation_list[i] maps to marker [i+1]
|
| 125 |
+
)
|
| 126 |
+
|
| 127 |
+
print(result["scores"])
|
| 128 |
+
```
|
| 129 |
+
|
| 130 |
## Data Collection And Use
|
| 131 |
|
| 132 |
RWGBench is built from public scholarly documents and metadata. Source
|
|
|
|
| 134 |
terms of the underlying papers when redistributing document-derived text.
|
| 135 |
|
| 136 |
The benchmark is intended for research on retrieval-augmented generation,
|
| 137 |
+
citation selection, scholarly writing evaluation, and related work generation.
|