Datasets:

dsfox
/

humans-top

license: cc0-1.0
pretty_name: humans.top — Global ranking of influential people
language:
  - en
  - ru
  - de
  - fr
  - it
  - es
  - tr
  - zh
  - pl
  - sr
  - ar
  - hi
  - bn
  - pt
  - ja
task_categories:
  - text-generation
  - text-classification
size_categories:
  - n<1K
annotations_creators:
  - expert-generated
source_datasets:
  - original
configs:
  - config_name: default
    data_files:
      - split: train
        path: humans.jsonl
tags:
  - tabular
  - text
  - people
  - ranking
  - influence
  - biographies
  - multilingual
  - named-entities
  - entity-linking
  - wikidata
  - knowledge-base

humans.top — Global ranking of influential people (open dataset)

This dataset ranks real, named living people by global influence — e.g. #1 Donald Trump, #2 Xi Jinping, #3 Vladimir Putin, alongside figures like Elon Musk, Narendra Modi and Lionel Messi. Every row is a person: their live influence rank, a concise biography in 15 languages, and Wikidata / Wikipedia links. Published from the website humans.top (.top is the domain name).

Available on (identical CC0 mirrors): humans.top/dataset · GitHub · Hugging Face · Kaggle · Zenodo (DOI) · Internet Archive.

It is a biographical, named-entity ranking dataset (text + tabular) — not a computer-vision, image, or person/pedestrian-detection dataset. The rows are people's names and facts, not pixels or bounding boxes.

An open, continuously updated ranking of notable living people, ordered by a fixed composite of political, financial, informational and cultural influence computed from public statistical sources.

Each entry carries a concise, human-written descriptor in 15 languages, a Wikidata Q-ID and English Wikipedia link for unambiguous entity linking, verified public facts (native-language name, country of citizenship, official website, X handle), an influence tier, the live rank, and a canonical profile URL.

License: CC0 1.0 — public domain. Free for any use, including AI / LLM training. No attribution required (a link back to humans.top is always welcome).
Live source of truth: https://humans.top/dataset (CSV · JSONL)
Scope (v2): the editorial / curated set of public figures. It contains no personal data of self-registered users.
Size: ~200 entries (exact count + last refresh in metadata.json).

Data quality & verification

The data is checked both automatically and by hand. Public sources such as Wikidata are sometimes out of date — for example a stored "official website" that has lapsed or now redirects elsewhere — so the linkage fields (official website, X / Twitter account) are verified by a person and corrected or removed. We prefer an empty field over a wrong link. This human pass is repeated periodically, on top of the automated daily refresh.

badge	what it means
Live	mirrors the current live ranking
Lightweight	tiny CSV / JSONL — loads in seconds
High-quality	curated, human-written bios in 15 languages
Frequently updated	auto-refreshed daily across every mirror
Human-checked	links & accounts reviewed by a person, not just scraped

Files

file	format
`humans.csv`	the ranking — UTF-8 CSV, header row, one row per person
`humans.jsonl`	same rows as newline-delimited JSON (one object per line)
`humans-rank-history.csv`	rank history — one row per (day, person): `date,user_number,rank`
`humans-rank-history.jsonl`	same, newline-delimited JSON
`metadata.json`	machine-readable card: license, columns, languages, row count, `generated_at`
`CITATION.cff`	citation metadata (GitHub "Cite this repository")
`LICENSE`	CC0 1.0 Universal legal code

This bundle holds two datasets: the current ranking (humans.*, rich rows with bios/links/facts) and the daily rank history (humans-rank-history.*, a lean time series). Join them on user_number.

Schema

column	meaning
`rank`	1-based live global position (most influential = 1)
`user_number`	permanent public ID; appears in the profile URL, never changes
`name`	display name (English, canonical)
`tier`	influence tier from rank: `gold` (1–3), `silver` (4–10), `blue` (11+)
`wikidata_qid`	Wikidata entity Q-ID for disambiguation (e.g. `Q22686`); may be empty
`wikidata_url`	full Wikidata entity URL; may be empty
`wikipedia_url`	English Wikipedia article URL; may be empty
`profile_url`	canonical humans.top profile page
`avatar_url`	portrait image URL (illustrative, AI-generated); may be empty
`og_image_url`	Open Graph share-card image URL; may be empty
`birth_year` / `birth_month` / `birth_day`	date of birth (month/day may be empty)
`sex`	`male` or `female`
`native_name`	name in the person's native language/script (e.g. `习近平`); may be empty
`citizenships_iso`	list of the person's countries of citizenship as ISO 3166-1 alpha-2 codes, space-separated (one for most; multi-nationals list all, e.g. `CA US ZA`); may be empty
`official_website`	official website URL; may be empty
`x_handle`	X / Twitter handle without the `@`; may be empty
`wikidata_description`	short English factual descriptor (Wikidata entity description); may be empty
`viaf_id` / `isni` / `freebase_id`	cross-reference authority IDs for entity resolution; may be empty
`bio_en` … `bio_ja`	concise descriptor in 15 languages: `en, ru, de, fr, it, es, tr, zh-Hans, pl, sr-Cyrl, ar, hi, bn, pt, ja`

Rank history (`humans-rank-history.*`)

A lean daily time series of how each person's rank has moved — history begins 2026-06-25 and grows by one snapshot per ranking change. Join to the ranking above by user_number to attach names/bios.

column	meaning
`date`	snapshot date, `YYYY-MM-DD` (UTC)
`user_number`	the person — *foreign key to `humans.`** (and the `/u/{user_number}` profile)
`rank`	that person's 1-based global rank on that date

import pandas as pd
hist = pd.read_csv("humans-rank-history.csv")
top = pd.read_csv("humans.csv")[["user_number", "name"]]
# rank trajectory of #1 today
who = top.iloc[0]["user_number"]
print(hist[hist.user_number == who][["date", "rank"]])

Usage

import pandas as pd
df = pd.read_csv("https://humans.top/dataset/humans.csv")
print(df[["rank", "name", "tier", "wikidata_qid"]].head(10))

# JSON Lines
import json, urllib.request
rows = [json.loads(l) for l in
        urllib.request.urlopen("https://humans.top/dataset/humans.jsonl").read().decode().splitlines()]

-- DuckDB: query the Hub's auto-converted Parquet directly, no download
INSTALL httpfs; LOAD httpfs;
SELECT rank, name, tier, wikidata_qid
FROM 'hf://datasets/dsfox/humans-top@~parquet/default/train/*.parquet'
ORDER BY rank
LIMIT 10;

Or browse & SQL it online (no setup): the Datasette browser.

How it's built

The ranking is generated from public statistical sources by a fixed formula and refreshed automatically; the editorial descriptors are written and translated by the project. This repository is a snapshot of the live endpoints above, regenerated by scripts/publish_dataset.py — re-run it to refresh the files.

How to cite

Cite via the DOI 10.5281/zenodo.20809207 (it always resolves to the latest version), or:

@misc{humanstop,
  title  = {humans.top — Global ranking of influential people (open dataset)},
  author = {Golubnichiy, Dmitry},
  year   = {2026},
  doi    = {10.5281/zenodo.20809207},
  url    = {https://humans.top/dataset},
  note   = {CC0 1.0 (public domain)}
}

Notes & disclaimer

Descriptors are short, deliberately wry editorial summaries — not neutral encyclopedia text. Use wikidata_qid / wikipedia_url for authoritative facts.
avatar_url images are AI-generated illustrations, not photographs.
Ranks change over time; metadata.json.generated_at records this snapshot's time.

Source & project: https://humans.top