Update README.md
Browse files
README.md
CHANGED
@@ -104,4 +104,65 @@ configs:
|
|
104 |
data_files:
|
105 |
- split: train
|
106 |
path: data/train-*
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
107 |
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
104 |
data_files:
|
105 |
- split: train
|
106 |
path: data/train-*
|
107 |
+
license: cc-by-4.0
|
108 |
+
task_categories:
|
109 |
+
- automatic-speech-recognition
|
110 |
+
language:
|
111 |
+
- fr
|
112 |
+
tags:
|
113 |
+
- medical
|
114 |
+
pretty_name: PxCorpus
|
115 |
+
size_categories:
|
116 |
+
- 1K<n<10K
|
117 |
---
|
118 |
+
|
119 |
+
# PxCorpus : A Spoken Drug Prescription Dataset in French
|
120 |
+
|
121 |
+
PxCorpus is to the best of our knowledge, the first spoken medical drug prescriptions corpus to be distributed.
|
122 |
+
It contains 4 hours of transcribed and annotated dialogues of drug prescriptions in
|
123 |
+
French acquired through an experiment with 55 participants experts and non-experts in drug prescriptions.
|
124 |
+
|
125 |
+
The automatic transcriptions were verified by human effort and aligned with
|
126 |
+
semantic labels to allow training of NLP models. The data acquisition protocol
|
127 |
+
was reviewed by medical experts and permit free distribution without breach of
|
128 |
+
privacy and regulation.
|
129 |
+
|
130 |
+
## Overview of the Corpus
|
131 |
+
The experiment has been performed in wild conditions with naive participants and medical experts.
|
132 |
+
In total, the dataset includes 2067 recordings of 55 participants (38% non-experts,
|
133 |
+
25% doctors, 36% medical practitioners), manually transcribed and semantically annotated.
|
134 |
+
|
135 |
+
| Category | Sessions | Recordings | Time(m)|
|
136 |
+
|------------------| -------- | ---------- | ------ |
|
137 |
+
| Medical experts | 258 | 434 | 94.83 |
|
138 |
+
| Doctors | 230 | 570 | 105.21 |
|
139 |
+
| Non experts | 415 | 977 | 62.13 |
|
140 |
+
| Total | 903 | 1981 | 262.27 |
|
141 |
+
|
142 |
+
## License
|
143 |
+
We hope that that the community will be able to benefit from the dataset
|
144 |
+
which is distributed with an attribution 4.0 International (CC BY 4.0) Creative Commons licence.
|
145 |
+
|
146 |
+
## How to cite this corpus
|
147 |
+
|
148 |
+
If you use the corpus or need more details please refer to the following paper: A spoken drug prescription datset in French for spoken Language Understanding
|
149 |
+
|
150 |
+
@InProceedings{Kocabiyikoglu2022,
|
151 |
+
author = "Alican Kocabiyikoglu and Fran{\c c}ois Portet and Prudence Gibert and Hervé Blanchon and Jean-Marc Babouchkine and Gaëtan Gavazzi",
|
152 |
+
title = "A spoken drug prescription datset in French for spoken Language Understanding",
|
153 |
+
booktitle = "13th Language Ressources and Evaluation Conference (LREC 2022)",
|
154 |
+
year = "2022",
|
155 |
+
location = "Marseille, France"
|
156 |
+
}
|
157 |
+
|
158 |
+
|
159 |
+
## Dataset features
|
160 |
+
|
161 |
+
* `path` -- Audio name
|
162 |
+
* `text` -- Audio utterance
|
163 |
+
* `ner` -- Semantic annotation from the original dataset
|
164 |
+
* `speaker_id` -- Speaker ID
|
165 |
+
* `speaker_age_range` -- Speaker age range
|
166 |
+
* `speaker_gender` -- Speaker gender
|
167 |
+
* `speaker_category` -- Speaker category (doctor, expert, non-expert)
|
168 |
+
* Other column names are for the occurences of each NER tag, could be useful for computing some metrics
|