0601p/Traveling_Namuwiki_Paths

Traveling Namuwiki

Traveling Namuwiki is a graph-navigation dataset built from Namuwiki page links. Each example contains a start page, a target page, and one or more valid paths between them. Paths are stored as intermediate page-title lists, excluding the start and target pages.

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

Files

data/train.jsonl
data/validation.jsonl
data/test.jsonl

Schema

Each JSONL row has this shape:

{
  "start_title": "source page title",
  "target_title": "destination page title",
  "paths": [["intermediate page title", "..."]],
  "hop": n
}

hop is the minimum number of hops needed to travel from start_title to target_title, and paths is the set of all shortest paths achieving that minimum. Each path lists only the intermediate page titles, so hop == len(paths[i]) + 1 for every i.

For a direct one-hop link, the path is an empty intermediate list:

{
  "start_title": "A",
  "target_title": "B",
  "paths": [[]],
  "hop": 1
}

Splits

The split was made with hop-aware stratification.

Split	Rows	Approx. ratio	File size
train	1,795,552	90.0%	1.91 GiB
validation	99,732	5.0%	107.2 MiB
test	99,767	5.0%	107.1 MiB
total	1,995,051	100.0%	2.11 GiB

Hop Distribution

Hop	train	validation	test
1	295,802	16,367	16,453
2	405,942	22,696	22,579
3	679,923	37,617	37,708
4	367,755	20,601	20,440
5	43,906	2,328	2,469
6	2,104	117	114
7	103	5	4
8	11	1	0
9	5	0	0
10	1	0	0

Usage

Load directly with datasets:

from datasets import load_dataset

data_files = {
    "train": "data/train.jsonl",
    "validation": "data/validation.jsonl",
    "test": "data/test.jsonl",
}

dataset = load_dataset("json", data_files=data_files)
print(dataset["train"][0])

or load with Hugging Face Hub:

from datasets import load_dataset

dataset = load_dataset("0601p/Traveling_namuwiki")

Notes

• Page titles come from the source Namuwiki dump and may preserve artifacts from the original dump encoding.
• paths contains intermediate pages only. Add start_title and target_title around a path if a full route is needed.