Datasets:
Tasks:
Text Generation
Modalities:
Text
Formats:
json
Languages:
English
Size:
10K - 100K
ArXiv:
License:
Update README.md
Browse files
README.md
CHANGED
|
@@ -52,24 +52,6 @@ You can scrape and clean all 16 guideline sources using our code in [epfLLM/medi
|
|
| 52 |
- **Paper:** *[MediTron-70B: Scaling Medical Pretraining for Large Language Models]()*
|
| 53 |
- **Knowledge Cutoff**: August 2023
|
| 54 |
|
| 55 |
-
|
| 56 |
-
## Dataset Structure
|
| 57 |
-
|
| 58 |
-
<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
|
| 59 |
-
|
| 60 |
-
Each row of the dataset represents one clinical practice guideline article, and consists of the following dataset fields (all strings):
|
| 61 |
-
|
| 62 |
-
|
| 63 |
-
| Field | Description | Sources with field |
|
| 64 |
-
|-------------|-------------------------------------------|------------------------------|
|
| 65 |
-
| `id` | Unique identifier for each article | All |
|
| 66 |
-
| `source` | Source tag (`cco`, `cdc`, `cma`, `icrc`, `nice`, `spor`, `who` or `wikidoc`)| All |
|
| 67 |
-
| `title` | Title of the article | CMA, NICE & WikiDoc only |
|
| 68 |
-
| `url` | URL of the article | NICE, WikiDoc only |
|
| 69 |
-
| `raw_text` | Unprocessed scraped article text | All |
|
| 70 |
-
| `clean_text`| Cleaned and formatted article text | All |
|
| 71 |
-
| `overview` | Short summary of the article | NICE only |
|
| 72 |
-
|
| 73 |
## Dataset Creation
|
| 74 |
|
| 75 |
### Curation Rationale
|
|
@@ -120,6 +102,24 @@ Finally, all samples were deduplicated using title matching, and articles that w
|
|
| 120 |
|
| 121 |
As the articles are publicly accessible, no personal or sensitive information is included.
|
| 122 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 123 |
## Uses
|
| 124 |
|
| 125 |
<!-- Address questions around how the dataset is intended to be used. -->
|
|
|
|
| 52 |
- **Paper:** *[MediTron-70B: Scaling Medical Pretraining for Large Language Models]()*
|
| 53 |
- **Knowledge Cutoff**: August 2023
|
| 54 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
## Dataset Creation
|
| 56 |
|
| 57 |
### Curation Rationale
|
|
|
|
| 102 |
|
| 103 |
As the articles are publicly accessible, no personal or sensitive information is included.
|
| 104 |
|
| 105 |
+
## Dataset Structure
|
| 106 |
+
|
| 107 |
+
<!-- This section provides a description of the dataset fields, and additional information about the dataset structure such as criteria used to create the splits, relationships between data points, etc. -->
|
| 108 |
+
|
| 109 |
+
Each row of the dataset represents one clinical practice guideline article, and consists of the following dataset fields (all strings):
|
| 110 |
+
|
| 111 |
+
|
| 112 |
+
| Field | Description | Sources with field |
|
| 113 |
+
|-------------|-------------------------------------------|------------------------------|
|
| 114 |
+
| `id` | Unique identifier for each article | All |
|
| 115 |
+
| `source` | Source tag (`cco`, `cdc`, `cma`, `icrc`, `nice`, `spor`, `who` or `wikidoc`)| All |
|
| 116 |
+
| `title` | Title of the article | CMA, NICE & WikiDoc only |
|
| 117 |
+
| `url` | URL of the article | NICE, WikiDoc only |
|
| 118 |
+
| `raw_text` | Unprocessed scraped article text | All |
|
| 119 |
+
| `clean_text`| Cleaned and formatted article text | All |
|
| 120 |
+
| `overview` | Short summary of the article | NICE only |
|
| 121 |
+
|
| 122 |
+
|
| 123 |
## Uses
|
| 124 |
|
| 125 |
<!-- Address questions around how the dataset is intended to be used. -->
|