20 news group (20ng)
Authors/Creators
Description
20 Newsgroups (20NG) is a classical and popular dataset for experiments in text applications of machine learning techniques. It contains 18,846 newsgroups documents, partitioned almost evenly across 20 different newsgroups categories.
http://qwone.com/~jason/20Newsgroups/
The files:
texts.txt: Document set (text). One per line.
score.txt: Document class whose index is associated with texts.txt
split_<k>.pkl: pandas DataFrame with k-cross validation partition.
The .zip contains all aforementioned files + the tfidf representation in the CSR matrix format.
Label Definition: (Score File)
0 atheist resources
1 computer graphics
2 computer os ms windows misc
3 computer system ibm pc hardware
4 computer system mac hardware
5 computer windows x
6 misc miscellaneous for sale
7 rec autos
8 rec motorcycles
9 rec sport baseball
10 rec sport hockey
11 science crypt
12 science electronics
13 science med
14 science space
15 society religion christian
16 talk politics guns
17 talk politics mideast
18 talk politics misc miscellaneous
19 talk religion misc miscellaneous
Files
20ng.zip
Files
(217.0 MB)
| Name | Size | Download all |
|---|---|---|
|
md5:9a16a9dad9e198150b3722f85049617c
|
186.8 MB | Preview Download |
|
md5:ed9f78d6e7ecae21a1e4771961d36731
|
47.1 kB | Preview Download |
|
md5:6dcb4e1cba6495b43b2c99f0f110f7df
|
564.2 kB | Download |
|
md5:89a3dcea452e734197f3629735634771
|
282.5 kB | Download |
|
md5:ca453fc9ee1f554ffd95af0cc73df509
|
29.3 MB | Preview Download |
Additional details
References
- https://doi.org/10.1016/j.ipm.2020.102263
- https://doi.org/10.1016/j.ipm.2020.102481
- Newsweeder: Learning to filter netnews