Link dataset to paper and GitHub repository, add Python API usage

#1
by nielsr HF Staff - opened
Files changed (1) hide show
  1. README.md +32 -6
README.md CHANGED
@@ -1,18 +1,18 @@
1
  ---
 
 
2
  license: mit
 
 
3
  task_categories:
4
  - text-generation
5
  - text-retrieval
6
- language:
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.