README.md

metadata

license: cc-by-nc-sa-4.0
task_categories:
  - text-generation
  - feature-extraction
language:
  - en
tags:
  - soulslike
  - game
  - rpg
  - dark-fantasy
  - lore
  - rag
  - knowledge-base
  - nlp
  - web-scraping
  - elden-ring
  - shadow-of-the-erdtree
  - nightreign
  - dark-souls
  - bloodborne
  - sekiro
  - demons-souls
  - fextralife
pretty_name: Soulslike Wiki Scrapping
size_categories:
  - 10K<n<100K

Soulslike Wiki Knowledge Base (RAG Corpus)

Dataset Description

This dataset contains structured text scraped from public wikis (primarily Fextralife) covering major "Soulslike" games developed by FromSoftware. It is designed to serve as a Knowledge Base for RAG (Retrieval-Augmented Generation) systems, fine-tuning LLMs, or game analytics.

The current release is stored as a hierarchical JSON where:

{
  "<Project/Game>": {
    "<PageSlug>": {
      "meta": {...},
      "data": {...}
    }
  }
}

• Curated content: Irrelevant pages (calculators, login screens, forums) have been filtered out.
• Format: JSON snapshots uploaded under FelipeRochaMartins/SoulsWikiScrapping (compatible with Hugging Face datasets).
• Games Covered: Elden Ring (including Shadow of the Erdtree & Nightreign), Dark Souls Trilogy, Bloodborne, Sekiro, Demon's Souls.

Dataset Structure

Data Fields

Each page entry is stored beneath its parent project using two top-level keys:

• meta: Metadata about the page.
  • url: Source URL.
  • project: Game identifier (e.g., Bloodborne).
• data: Scraped content payload.
  • title: Human-readable title (e.g., "1st Floor Sickroom | Bloodborne Wiki").
  • content: Raw text extracted from the wiki body.

Note: In the JSON snapshot the keys inside each project omit the .json suffix (e.g., "1st_Floor_Sickroom"). When importing back to data/raw, the loader automatically restores the .json extension.

Example Instance

{
  "Bloodborne": {
    "1st_Floor_Sickroom": {
      "meta": {
        "url": "https://bloodborne.wiki.fextralife.com/1st+Floor+Sickroom",
        "project": "Bloodborne"
      },
      "data": {
        "title": "1st Floor Sickroom | Bloodborne Wiki",
        "content": "1st Floor Sickroom is the tutorial area of Bloodborne..."
      }
    }
  }
}

Usage

How to load in Python

from datasets import load_dataset

dataset = load_dataset("FelipeRochaMartins/SoulsWikiScrapping", data_files="latest.json")
payload = dataset["train"][0]

# Access a project + page pair
bloodborne = payload["Bloodborne"]
first_page = bloodborne["1st_Floor_Sickroom"]
print(first_page["data"]["title"])

Source & Licensing

Source: The data was scraped from Fextralife Wiki. License: This dataset is distributed under the CC BY-NC-SA 4.0 license (Creative Commons Attribution-NonCommercial-ShareAlike), consistent with the licensing terms of the source wikis.

• Attribution: All narrative content belongs to FromSoftware and the respective wiki contributors.
• Non-Commercial: This dataset should not be used for commercial purposes without permission from the copyright holders.

Intended Use

• RAG Systems: Building a chatbot that answers questions about Elden Ring/Bloodborne/Dark Souls/Sekiro/Demon's Souls lore.
• Lore Analysis: Analyzing entity relationships (e.g., "Which NPCs are related to Miquella?").
• Style Transfer: Training models to write in the archaic/fantasy style of Hidetaka Miyazaki.