Datasets:

Modalities:
Text
Formats:
json
ArXiv:
Libraries:
Datasets
pandas
License:
tonytan48 commited on
Commit
e0ab348
1 Parent(s): e0521ab

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +36 -0
README.md CHANGED
@@ -1,3 +1,39 @@
1
  ---
2
  license: mit
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: mit
3
  ---
4
+ # Re-DocRED Dataset
5
+
6
+ This repository contains the dataset of our EMNLP 2022 research paper [Revisiting DocRED – Addressing the False Negative Problem
7
+ in Relation Extraction](https://arxiv.org/pdf/2205.12696.pdf).
8
+
9
+ DocRED is a widely used benchmark for document-level relation extraction. However, the DocRED dataset contains a significant percentage of false negative examples (incomplete annotation). We revised 4,053 documents in the DocRED dataset and resolved its problems. We released this dataset as: Re-DocRED dataset.
10
+
11
+ The Re-DocRED Dataset resolved the following problems of DocRED:
12
+ 1. Resolved the incompleteness problem by supplementing large amounts of relation triples.
13
+ 2. Addressed the logical inconsistencies in DocRED.
14
+ 3. Corrected the coreferential errors within DocRED.
15
+
16
+
17
+ # Statistics of Re-DocRED
18
+ The Re-DocRED dataset is located as ./data directory, the statistics of the dataset are shown below:
19
+
20
+
21
+ | | Train | Dev | Test |
22
+ | :---: | :-: | :-: |:-: |
23
+ | # Documents | 3,053 | 500 | 500 |
24
+ | Avg. # Triples | 28.1 | 34.6 | 34.9 |
25
+ | Avg. # Entities | 19.4 | 19.4 | 19.6 |
26
+ | Avg. # Sents | 7.9 | 8.2 | 7.9 |
27
+
28
+ # Citation
29
+ If you find our work useful, please cite our work as:
30
+ ```bibtex
31
+ @inproceedings{tan2022revisiting,
32
+ title={Revisiting DocRED – Addressing the False Negative Problem in Relation Extraction},
33
+ author={Tan, Qingyu and Xu, Lu and Bing, Lidong and Ng, Hwee Tou and Aljunied, Sharifah Mahani},
34
+ booktitle={Proceedings of EMNLP},
35
+ url={https://arxiv.org/abs/2205.12696},
36
+ year={2022}
37
+ }
38
+ ```
39
+