---
task_categories:
- token-classification
- text-classification
language:
- en
tags:
- wikipedia
- english
pretty_name: WIKI 727k
size_categories:
- 100K<n<1M
---


The original dataset repo url : [https://github.com/koomri/text-segmentation](https://github.com/koomri/text-segmentation)

for every wiki article text we use the following code to produce label and input text.

```python
label = []
text = ""
for line in data:
    if line.startswith("========"):
        if len(label) == 0:
            continue
        else:
            label[-1] = 1
    else:
        text += line + "\n"
        label.append(0)
```