Datasets:
Tasks:
Text Classification
Modalities:
Text
Formats:
json
Languages:
English
Size:
10K - 100K
License:
sagnikrayc
commited on
Commit
•
fba1336
1
Parent(s):
db9acfa
Update README.md
Browse files
README.md
CHANGED
@@ -39,6 +39,25 @@ Moreover, SpanEx is annotated by three annotators, which opens new avenues for s
|
|
39 |
Our study reveals that while human annotators often agree on span interactions, they also offer complementary reasons for a prediction, collectively providing a comprehensive set of reasons for a prediction.
|
40 |
We collect explanations of span interactions for NLI on the SNLI dataset and for FC on the FEVER dataset.
|
41 |
|
42 |
-
|
43 |
|
44 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
Our study reveals that while human annotators often agree on span interactions, they also offer complementary reasons for a prediction, collectively providing a comprehensive set of reasons for a prediction.
|
40 |
We collect explanations of span interactions for NLI on the SNLI dataset and for FC on the FEVER dataset.
|
41 |
|
42 |
+
Please cite the following paper if you use this dataset:
|
43 |
|
44 |
+
```
|
45 |
+
@inproceedings{choudhury-etal-2023-explaining,
|
46 |
+
title = "Explaining Interactions Between Text Spans",
|
47 |
+
author = "Choudhury, Sagnik and
|
48 |
+
Atanasova, Pepa and
|
49 |
+
Augenstein, Isabelle",
|
50 |
+
editor = "Bouamor, Houda and
|
51 |
+
Pino, Juan and
|
52 |
+
Bali, Kalika",
|
53 |
+
booktitle = "Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing",
|
54 |
+
month = dec,
|
55 |
+
year = "2023",
|
56 |
+
address = "Singapore",
|
57 |
+
publisher = "Association for Computational Linguistics",
|
58 |
+
url = "https://aclanthology.org/2023.emnlp-main.783",
|
59 |
+
doi = "10.18653/v1/2023.emnlp-main.783",
|
60 |
+
pages = "12709--12730",
|
61 |
+
abstract = "Reasoning over spans of tokens from different parts of the input is essential for natural language understanding (NLU) tasks such as fact-checking (FC), machine reading comprehension (MRC) or natural language inference (NLI). However, existing highlight-based explanations primarily focus on identifying individual important features or interactions only between adjacent tokens or tuples of tokens. Most notably, there is a lack of annotations capturing the human decision-making process with respect to the necessary interactions for informed decision-making in such tasks. To bridge this gap, we introduce SpanEx, a multi-annotator dataset of human span interaction explanations for two NLU tasks: NLI and FC. We then investigate the decision-making processes of multiple fine-tuned large language models in terms of the employed connections between spans in separate parts of the input and compare them to the human reasoning processes. Finally, we present a novel community detection based unsupervised method to extract such interaction explanations. We make the code and the dataset available on [Github](https://github.com/copenlu/spanex). The dataset is also available on [Huggingface datasets](https://huggingface.co/datasets/copenlu/spanex).",
|
62 |
+
}
|
63 |
+
```
|