Commit
·
c7539f9
1
Parent(s):
b4d7d80
rollback changes
Browse files
README.md
CHANGED
|
@@ -56,14 +56,19 @@ task_ids:
|
|
| 56 |
|
| 57 |
### Dataset Summary
|
| 58 |
|
| 59 |
-
The `tldr_news` dataset was constructed by collecting a daily tech newsletter (
|
| 60 |
-
|
| 61 |
-
content
|
|
|
|
| 62 |
Such a dataset can be used to train a model to generate a headline from a input piece of text.
|
| 63 |
|
| 64 |
### Supported Tasks and Leaderboards
|
| 65 |
|
| 66 |
-
There is no official supported tasks nor leaderboard for this dataset.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 67 |
|
| 68 |
### Languages
|
| 69 |
|
|
@@ -90,22 +95,31 @@ An example is as follows:
|
|
| 90 |
|
| 91 |
### Data Splits
|
| 92 |
|
| 93 |
-
|
| 94 |
|
| 95 |
## Dataset Creation
|
| 96 |
|
| 97 |
### Curation Rationale
|
| 98 |
|
| 99 |
This dataset was obtained by scrapping the collecting all the existing newsletter
|
| 100 |
-
available [here](https://tldr.tech/newsletter).
|
| 101 |
-
|
|
|
|
|
|
|
| 102 |
|
| 103 |
### Source Data
|
| 104 |
|
| 105 |
#### Initial Data Collection and Normalization
|
| 106 |
|
| 107 |
-
The dataset was has been collected from https://tldr.tech/newsletter.
|
| 108 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 109 |
|
| 110 |
#### Who are the source language producers?
|
| 111 |
|
|
@@ -115,12 +129,12 @@ The people (or person) behind the https://tldr.tech/ newsletter.
|
|
| 115 |
|
| 116 |
#### Annotation process
|
| 117 |
|
| 118 |
-
|
|
|
|
| 119 |
|
| 120 |
#### Who are the annotators?
|
| 121 |
|
| 122 |
-
|
| 123 |
-
used as such.
|
| 124 |
|
| 125 |
### Personal and Sensitive Information
|
| 126 |
|
|
@@ -134,17 +148,17 @@ used as such.
|
|
| 134 |
|
| 135 |
### Discussion of Biases
|
| 136 |
|
| 137 |
-
|
| 138 |
|
| 139 |
### Other Known Limitations
|
| 140 |
|
| 141 |
-
|
| 142 |
|
| 143 |
## Additional Information
|
| 144 |
|
| 145 |
### Dataset Curators
|
| 146 |
|
| 147 |
-
The dataset was obtained by
|
| 148 |
|
| 149 |
### Contributions
|
| 150 |
|
|
|
|
| 56 |
|
| 57 |
### Dataset Summary
|
| 58 |
|
| 59 |
+
The `tldr_news` dataset was constructed by collecting a daily tech newsletter (available
|
| 60 |
+
[here](https://tldr.tech/newsletter)). Then, for every piece of news, the `headline` and its corresponding `
|
| 61 |
+
content` were extracted.
|
| 62 |
+
|
| 63 |
Such a dataset can be used to train a model to generate a headline from a input piece of text.
|
| 64 |
|
| 65 |
### Supported Tasks and Leaderboards
|
| 66 |
|
| 67 |
+
There is no official supported tasks nor leaderboard for this dataset. However, it could be used for the following
|
| 68 |
+
tasks:
|
| 69 |
+
|
| 70 |
+
- summarization
|
| 71 |
+
- headline generation
|
| 72 |
|
| 73 |
### Languages
|
| 74 |
|
|
|
|
| 95 |
|
| 96 |
### Data Splits
|
| 97 |
|
| 98 |
+
- `all`: all existing daily newsletters available [here](https://tldr.tech/newsletter).
|
| 99 |
|
| 100 |
## Dataset Creation
|
| 101 |
|
| 102 |
### Curation Rationale
|
| 103 |
|
| 104 |
This dataset was obtained by scrapping the collecting all the existing newsletter
|
| 105 |
+
available [here](https://tldr.tech/newsletter).
|
| 106 |
+
|
| 107 |
+
Every single newsletter was then processed to extract all the different pieces of news. Then for every collected piece
|
| 108 |
+
of news the headline and the news content were extracted.
|
| 109 |
|
| 110 |
### Source Data
|
| 111 |
|
| 112 |
#### Initial Data Collection and Normalization
|
| 113 |
|
| 114 |
+
The dataset was has been collected from https://tldr.tech/newsletter.
|
| 115 |
+
|
| 116 |
+
In order to clean up the samples and to construct a dataset better suited for headline generation we have applied a
|
| 117 |
+
couple of normalization steps:
|
| 118 |
+
|
| 119 |
+
1. The headlines initially contain an estimated read time in parentheses; we stripped this information from the
|
| 120 |
+
headline.
|
| 121 |
+
2. Some headlines are just a technology/repo/software name. We filtered out such samples if this name was not mention in
|
| 122 |
+
the news content.
|
| 123 |
|
| 124 |
#### Who are the source language producers?
|
| 125 |
|
|
|
|
| 129 |
|
| 130 |
#### Annotation process
|
| 131 |
|
| 132 |
+
Disclaimers: The dataset was generated from a daily newsletter. The author had no intention for those newsletters to be
|
| 133 |
+
used as such.
|
| 134 |
|
| 135 |
#### Who are the annotators?
|
| 136 |
|
| 137 |
+
The newsletters were written by the people behind *TLDR tech*.
|
|
|
|
| 138 |
|
| 139 |
### Personal and Sensitive Information
|
| 140 |
|
|
|
|
| 148 |
|
| 149 |
### Discussion of Biases
|
| 150 |
|
| 151 |
+
This dataset only contains tech news. A model trained on such a dataset might not be able to generalize to other domain.
|
| 152 |
|
| 153 |
### Other Known Limitations
|
| 154 |
|
| 155 |
+
[Needs More Information]
|
| 156 |
|
| 157 |
## Additional Information
|
| 158 |
|
| 159 |
### Dataset Curators
|
| 160 |
|
| 161 |
+
The dataset was obtained by collecting newsletters from this website: https://tldr.tech/newsletter
|
| 162 |
|
| 163 |
### Contributions
|
| 164 |
|