The Agency Against Corruption (AAC) of the Ministry of Justice convenes the Government Procurement Integrity Platform (機關採購廉政平臺), through which procuring entities, oversight bodies and the AAC jointly monitor the integrity of major public construction projects. As part of the platform’s digital advancement program, the AAC publishes OCDS data on the platform’s portal.
The publication covers major construction projects monitored under the Government Procurement Integrity Platform, rather than all public procurement in Taiwan. Procuring entities include state-owned enterprises (Taiwan Water Corporation, CPC Corporation, Taoyuan International Airport Corporation), agencies of the Ministry of Transportation and Communications (Railway Bureau and its Northern Engineering Office, Civil Aeronautics Administration, Highway Bureau), the Central Region Water Resources Bureau of the Ministry of Economic Affairs, the New Taipei City Government Procurement Office and the Taitung County Government.
A contracting process has several stages: tendering, awarding, contracting and implementation. You can learn about the contracting stages in this tutorial.
These counts are the number of JSON objects across all contracting processes. Parties, tenderers and suppliers are unique within a contracting process, but can be repeated across contracting processes. You can learn about data publication in this tutorial.
Parties
Count of parties:
0
Planning
Count of planning activities:
0
Tenders
Count of tenders:
154
Count of tenderers:
0
Count of tender items:
0
Awards
Count of awards:
158
Count of suppliers:
213
Count of award items:
0
Contracts
Count of contracts:
158
Count of contract items:
0
Count of transactions:
0
Documents
Count of documents:
0
Milestones
Count of milestones:
0
Amendments
Count of amendments:
0
No extensions are used. There are additional fields: tender/awardMethod (the award criteria), tender/bidOpeningDate (the date of the bid opening), tender/numberOfTenders (the number of tenderers), tender/tenderMethod (the procurement method), award/unsuccessfulTenderer/name and award/unsuccessfulTenderer/bidAmount (the names and bid values of unsuccessful tenderers), awards/suppliers/bidAmount (the winning bid value) and gainAmount.
Data quality
The dataset has quality issues to take into account, notably:
The procurement method is published in the additional tender/tenderMethod field instead of the standard tender/procurementMethod field, so analyses that filter by procurement method cannot rely on the standard field.
Information about tenderers is split across additional fields: the number of tenderers is in tender/numberOfTenders rather than tender/numberOfTenderers, and unsuccessful tenderers appear under award/unsuccessfulTenderer rather than in tender/tenderers, so analyses of participation and competition need to read these non-standard fields.
Bid values (award/unsuccessfulTenderer/bidAmount, awards/suppliers/bidAmount, gainAmount) are published in additional fields rather than using the bids extension, and the bid opening date and award criteria are in tender/bidOpeningDate and tender/awardMethod rather than the bid opening extension and tender/awardCriteriaDetails.
Last reviewed: May 2026
Access data
This OCDS dataset is available for download in JSON, Excel or CSV format. You can download the data for contracting processes in a specific year or for all time.
JSON
Each contracting process is represented as one line of JSON text in the .jsonl file.
The .jsonl file is compressed using Gzip. Windows users need 7-Zip, WinRAR or WinZip to decompress the .gz file.
Each contracting process is represented as one row in the main sheet. Other sheets link to it via the main_ocid column. In general, each sheet is linked to its parent sheet via the [sheet_name]_id column.
Excel files can have at most 1,048,576 rows. If a year is missing, it had too many rows. Download the CSV files, instead.
Each contracting process is represented as one row in the main.csv file. Other files link to it via the main_ocid column. In general, each file is link to its parent file via the [file_name]_id column.
The .csv files are archived using tar and compressed using Gzip. Windows users need 7-Zip, WinRAR or WinZip to decompress the .tar.gz file.