Hugging Face's logo

Datasets:
loreatec
/
japan-procurement-open-data

Tasks:
Tabular Classification
Table Question Answering
Text Retrieval
Modalities:
Tabular
Text
Formats:
csv
Languages:
Japanese
English
Size:
1M - 10M
Tags:
japan
public-procurement
government-contracts
open-data
subsidies
companies
Libraries:
Datasets
Dask
Polars
License:
Dataset card Data Studio Files
xet
 Community
1
japan-procurement-open-data / README.md
loreatec's picture
loreatec
Dataset card: sources, licences and caveats
be555f3 verified
preview code
|
Raw
History Blame Contribute Delete
7.05 kB
metadata 
license: other
license_name: mixed-japanese-government-open-data
license_link: LICENSE
language:
  - ja
  - en
tags:
  - japan
  - public-procurement
  - government-contracts
  - open-data
  - subsidies
  - companies
  - tenders
pretty_name: Japan Public Procurement & Company Open Data (LoreaTec)
size_categories:
  - 1M<n<10M
task_categories:
  - tabular-classification
  - table-question-answering
  - text-retrieval
configs:
  - config_name: awards
    data_files: awards_fy*.csv.gz
  - config_name: award_series
    data_files: award_series.csv.gz
  - config_name: awards_with_series
    data_files: awards_with_series.csv.gz
  - config_name: subsidies
    data_files: subsidies_gbizinfo.csv.gz
  - config_name: licenses
    data_files: licenses_gbizinfo.csv.gz
  - config_name: programme_payees
    data_files: rs_payees.csv.gz
  - config_name: programme_contracts
    data_files: rs_contracts.csv.gz
  - config_name: company_events
    data_files: company_events_90d.csv.gz

Japan Public Procurement & Company Open Data

Machine-readable extracts of Japanese public-procurement and company open data, compiled and normalised by LoreaTec for japan-tenders.loreatec.jp and bizsearch.loreatec.jp. Everything here comes from official Japanese government sources; the value added is the cleaning, joining and the derived analysis (contract series and re-tender predictions).

Updated monthly. The authoritative, always-current copy is https://bizsearch.loreatec.jp/datasets (with manifest.json).

What is inside

File Rows What it is Source
awards_fy2013…fy2026.csv.gz ~310,000 Every national-government (GEPS) contract award since FY2013: case number, title (ja), date, price (JPY), ministry code, procurement method, supplier name and corporate number 調達ポータル 落札実績オープンデータ
award_series.csv.gz ~42,000 Derived: contracts that repeat (same ministry + same normalised title), with cadence, typical month, last supplier/price, price statistics and the next expected round LoreaTec analysis of the above
awards_with_series.csv.gz ~200,000 The awards that belong to a series, with the normalised title and series id (lets you reproduce the grouping) LoreaTec analysis
subsidies_gbizinfo.csv.gz ~546,000 Subsidies, grants and transfers awarded to companies: corporate number, date, programme, amount, ministry gBizINFO 補助金情報 (METI)
licenses_gbizinfo.csv.gz ~130,000 Notifications, licences and certifications per company gBizINFO 届出・認定情報 (METI)
rs_payees.csv.gz ~149,000 Payees reported by each national budget programme (contractors, grantees, commissioned bodies) with corporate number and amount 行政事業レビュー見える化サイト
rs_contracts.csv.gz ~197,000 Individual contracts reported in those review sheets: method, number of bidders, winning ratio, amount 行政事業レビュー見える化サイト
company_events_90d.csv.gz varies Corporate-registry changes of the last 90 days: new companies, name changes, relocations, closures, mergers 国税庁法人番号公表サイト (daily diffs)
bidder_registration_windows.json 67 governments When each government (national + 47 prefectures + 20 designated cities) opens its bidder-registration window, with validity, method and official URL Curated by LoreaTec from official pages
forecast_predictions.json.gz ~13,000 Current radar of contracts expected to be re-tendered in the coming months LoreaTec analysis

Format: CSV, UTF-8 with BOM, CRLF, comma-separated, gzip-compressed. Amounts are in yen. Corporate numbers are the 13-digit 法人番号, which is the join key across every file (and against the NTA registry).

Licences — read this before redistributing

This repository mixes several Japanese government sources, each with its own terms. There is no single licence, which is why the repository is marked other:

  • 調達ポータル (Procurement Portal) award data — open data, reuse allowed with the attribution 「出典：調達ポータル（https://www.p-portal.go.jp/）」.
  • 行政事業レビュー見える化サイト — Public Data License 1.0 (PDL 1.0), compatible with CC BY 4.0, commercial use allowed. Requires 「出典：行政事業レビュー見える化サイト」 and a separate note that the data was edited (this repository is an edited extract).
  • gBizINFO (METI) — reuse allowed with attribution 「出典：gBizINFO（経済産業省）」.
  • 国税庁法人番号公表サイト (NTA corporate-number registry) — free reuse, attribution recommended.
  • LoreaTec's own analysis (series, predictions, normalisation, registration windows) — CC BY 4.0.

If you use any file, keep the attribution of its source column in the table above, state that the data was processed by LoreaTec, and do not present it as the government's original publication. Deliberately not included here: the full 5.8M-company registry (get it from the NTA), the prefectural construction-licence rosters (per-prefecture terms, some forbid redistribution), and anything personal.

Caveats

  • Award prices are whole-contract amounts; in multi-winner awards the price may be per unit. Quantities and durations differ, so read the contract titles before comparing.
  • Programme payees include intermediaries (agencies, funds, prefectures) that pass money on: do not sum naively across programmes.
  • award_series and forecast_predictions are statistical, not announcements. A prediction is "this ministry has re-tendered this contract every year since 2019", not a notice.
  • Corporate numbers are missing in a small share of rows (joint ventures, individuals, foreign bodies); 9999999999999 is the placeholder used by the source in the review sheets.
  • Company events start on 2023-08-16 (before that only each company's latest registry change is known).

Using it 

import pandas as pd
awards = pd.read_csv("hf://datasets/loreatec/japan-procurement-open-data/awards_fy2025.csv.gz")
subsidies = pd.read_csv("hf://datasets/loreatec/japan-procurement-open-data/subsidies_gbizinfo.csv.gz")
# join on the 13-digit corporate number
merged = awards.merge(subsidies, left_on="houjin_bangou", right_on="houjin_bangou", how="left")

An MCP server (for Claude, ChatGPT and other assistants) and plain JSON APIs over the same data live at https://bizsearch.loreatec.jp/mcp and https://bizsearch.loreatec.jp/data-api.

Citation 

LoreaTec (2026). Japan Public Procurement & Company Open Data.
Compiled from 調達ポータル, gBizINFO, 行政事業レビュー見える化サイト and 国税庁法人番号公表サイト.
https://bizsearch.loreatec.jp/datasets

Questions, corrections or a bigger API quota: info@loreatec.jp