0601p/Traveling_Namuwiki_Actions

Traveling Namuwiki Actions

Traveling Namuwiki Actions is an adjacency-list dataset built from Namuwiki page links. Each row contains a page title and the list of linked page titles that can be used as next actions in a graph-navigation task.

This dataset was derived from the Hugging Face dataset `heegyu/namuwiki`.

Files

data.jsonl

File size: about 405.1 MiB.

Schema

Each JSONL row has this shape:

{
  "title": "source page title",
  "actions": ["linked page title", "..."]
}

actions is sorted and deduplicated within each row. Pages with no extracted outgoing actions were removed from this dataset. Therefore, if a Namuwiki page title does not appear as a title row here, it should be treated as having no available outgoing actions in this action graph.

Statistics

Actions Per Page

Usage

Load the JSONL file directly:

import json

with open("data.jsonl", "r", encoding="utf-8") as f:
    for line in f:
        row = json.loads(line)
        title = row["title"]
        actions = row["actions"]
        break

Load with datasets:

from datasets import load_dataset

dataset = load_dataset("json", data_files="data.jsonl", split="train")
print(dataset[0])

or load with Hugging Face Hub:

from datasets import load_dataset

dataset = load_dataset("YOUR_USERNAME/Traveling_Namuwiki_Actions", split="train")

Build an in-memory graph:

graph = {row["title"]: row["actions"] for row in dataset}

Notes

• —This dataset stores page-title links only, not full article text.
• —Links to files, categories, images, external URLs, and self-links were removed during preprocessing.
• —Pages with no extracted outgoing links were removed. When traversing this graph, a title that is not present as a source title row should be handled as a terminal page with no outgoing actions.
• —Some linked page titles may not appear as source title rows due to redirects or missing pages; these should also be handled as terminal pages in this graph.