README.md

---
license: cc-by-4.0
language:
- en
configs:
- config_name: default
  data_files:
  - split: train
    path: '*.parquet'
size_categories:
- 100K<n<1M
---

# Roleplaying forum scrapes (raw)

Here are _mostly_ **original/raw files** for some of the roleplaying forums I scraped in the past (and some newly scraped ones), **repacked as HTML strings + some metadata on a one-row-per-thread basis** instead of a _one-row-per-message basis_, which should make them more convenient to handle.

Unlike [the previously uploaded archive](https://huggingface.co/datasets/lemonilia/Roleplay-Forums_2023-04), they shouldn't have issues with spaces between adjacent HTML tags, as that occurred by mistake in an intermediate processing step where single messages were extracted from the pages. Unfortunately, I do not have anymore the original files for most of the RP forums scraped in 2023; those would need to be scraped again.

## Scraping details
Most forums have been scraped page-by-page using the Firefox extension [Web Scraper](https://addons.mozilla.org/en-US/firefox/addon/web-scraper/), generally only picking the top-level thread message container instead of the entire page. The scraper was configured to:

1. Scan forum pages one by one, retrieve thread links and visit the links one by one;
2. Retrieve the thread page HTML, look for the "next page" link, navigate to that;
3. Repeat step 2 in a cyclic fashion.

For the process to work successfully, it was important not to scrape too many threads per run and using Firefox instead of Chrome, otherwise it would easily fail. Newer versions of Web Scraper solved some reliability issues with Firefox and made exporting the data to `.csv` format much quicker.

Using Python (pandas) on the exported data, pages from the same threads were grouped together and concatenated with the string `\n<!-- [NEW PAGE STARTS BELOW] -->\n`, without any further processing, obtaining one row of data per thread. The records were then saved into Parquet files (`compression='zstd', compression_level=7, row_group_size=20`).

I had to make sure to configure the row group size to a small value to avoid issues when loading the the files with pandas or in the HuggingFace dataset viewer (which supports a maximum row group size of 286 MB). As some of the RP threads are huge (more than 10 megabytes of size), I ultimately settled on 20, corresponding to about 200 MB in the worst case scenario.

## Usage notes
The data is not directly usable for finetuning as-is. You will need to split the threads into messages, extract metadata, clean/convert the HTML, etc. For the NSFW forums it will be probably best to somehow avoid using usernames directly, just to be nice(r).

### OOC
The dataset includes both in-character (IC) and out-of-character (OOC) threads. However, these are often not directly distinguishable except by tags in their titles. I strongly recommend filtering out OOC threads: OOC discussions often reference IC events without providing necessary background information, which can disrupt model coherence.

## Forum listing
Only the roleplay sections were scraped. Note that the SFW forums can have censored slurs, for example using asterisks like `****`.

| Name | Type | Scraping period | Notes
|------|------|-------|------
| AllTheFallen | NSFW (Lolisho) | 2024-10 |
| [Blue Moon Roleplaying Forums](https://bluemoonroleplaying.com/community/) | NSFW | 2023-03 |
| [Elliquiy](https://elliquiy.com/forums/index.php) | NSFW | 2023-03 to 2023-04 | "Print" version, lacks some formatting
| [Giant in the Playground](https://forums.giantitp.com/forumdisplay.php?3-Play-by-Post-Games) | SFW | 2025-01
| Lolicit | NSFW (Lolisho) | 2023-03 | Defunct website
| [Menewsha](https://www.menewsha.com/forum/forumdisplay.php?f=309) | SFW | 2025-01 | No activity at the time of scraping. Generally low-quality postings.
| [RPGnet Forums](https://forum.rpg.net/index.php?forums/roleplay-by-post-play-forum.31/) | SFW | 2025-01 | Both IC and OOC threads included

## Future plans
I might add more SFW forums in the future. For NSFW, what I included probably already covers most bases. Every forum needs its own cleaning process, so adding a large number of small forums, unless needed, isn't very time-efficient.