Please note: The algorithm descriptions in English have been automatically translated. Errors may have been introduced in this process. For the original descriptions, go to the Dutch version of the Algorithm Register.

Document search

Document experts check whether source documents, such as a birth certificate and marriage certificate, are genuine. The use of recognition algorithms supports document experts in document research.
Last change on 27th of May 2026, at 9:48 (CET) | Publication Standard 1.0
Publication category
Other algorithms
Impact assessment
DPIA
Status
In use

General information

Theme

Migration and Integration

Begin date

04-2024

Contact information

algoritmeregister@ind.nl

Link to publication website

https://ind.nl/nl/over-ons/onze-organisatie/bureau-documenten https://ind.nl/nl/over-ons/onze-organisatie/visualdoc

Responsible use

Goal and impact

The purpose of the algorithms is to process source documents faster. The document recognition function categorises source documents. This determines whether a source document is on a list drawn up by a document expert. This is a list of documents that are not investigated further due to low risk of fraud. The "detail recognition" function helps to find features of authenticity on source documents.

Considerations

Document experts face a very high workload and limited capacity. The algorithms used help document experts deal efficiently with source documents with a low risk profile.

Human intervention

The document recognition algorithm learns from how document experts categorise scanned source documents. The algorithm's categorisation is always checked manually.

Risk management

Document experts check the results of the document recognition algorithm. They assess whether the source document has been recognised correctly. If the algorithm has given an incorrect result, the document experts correct it manually. Source documents with a low risk profile are also examined manually at random. The document experts are always responsible for the final recommendation. Detail recognition is only carried out manually by document experts. The results of this algorithm are supportive and are not automatically used in further document examination.

Legal basis

  1. Articles 14, 20, 28 and 45a of the Aliens Act stipulate that the Minister is authorised to grant, amend and/or revoke residence permits.
  2. Article 3:2 of the General Administrative Law Act stipulates that when preparing a decision, the administrative body must gather the necessary knowledge regarding the relevant facts and interests to be weighed up. An obligation to conduct an investigation, therefore.

Links to legal bases

  • Article 14 Aliens Act: https://wetten.overheid.nl/jci1.3:c:BWBR0011823&hoofdstuk=3&afdeling=3&paragraaf=1&artikel=14&z=2024-01-01&g=2024-01-01
  • Article 20 Aliens Act: https://wetten.overheid.nl/jci1.3:c:BWBR0011823&hoofdstuk=3&afdeling=3&paragraaf=2&artikel=20&z=2024-01-01&g=2024-01-01
  • Article 28 Aliens Act: https://wetten.overheid.nl/jci1.3:c:BWBR0011823&hoofdstuk=3&afdeling=4&paragraaf=1&artikel=28&z=2024-01-01&g=2024-01-01
  • Section 45a Aliens Act: https://wetten.overheid.nl/jci1.3:c:BWBR0011823&hoofdstuk=3&afdeling=5&paragraaf=1&artikel=45a&z=2024-01-01&g=2024-01-01
  • - Article 3:2 General Administrative Law Act: https://wetten.overheid.nl/jci1.3:c:BWBR0005537&hoofdstuk=3&afdeling=3.2&artikel=3:2&z=2024-05-01&g=2024-05-01

Elaboration on impact assessments

DPIA has been carried out

Impact assessment

Data Protection Impact Assessment (DPIA)

Operations

Data

High-quality digital scans of source documents and human annotations, such as country, document type and location of authenticity features. Document experts review the algorithms' predictions and create new annotations. After checking these annotations, they are added to the dataset used by the algorithms. Thus, the algorithms continue to learn about patterns in the data based on feedback from the document experts.

Technical design

For the document recognition algorithm, a VGG16 neural network was trained on characteristics of document data based on source document data. This allows, for example, to recognise the document type. The detail recognition algorithm uses a combination of existing visual models and neural networks trained internally on authenticity characteristics.

The source documents are scanned and presented to the system. The scan is also presented to the document recognition algorithm. The predicted categorisation is then checked by a document expert before the source document continues in the document search. During this document search, authenticity characteristics can be manually retrieved using the detail recognition algorithm.

External provider

TNO

Similar algorithm descriptions

  • Automated document check and facial comparison for registrations in the BRP, applications for travel documents and driving licences, and for reporting removals (intra-municipal and migrations). The tool checks the authenticity characteristics of identity documents and prevents "look-alike fraud".
    Last change on 1st of May 2025, at 12:11 (CET) | Publication Standard 1.0
    Publication category
    Impactful algorithms
    Impact assessment
    Field not filled in.
    Status
    In use
  • At the civil affairs counter, old documents handed in for replacement for a new, valid one must be scanned for authenticity. In addition, identity documents are scanned for authenticity during identity checks. The document scanner scans the authenticity characteristics of a driving licence, identity card or passport.
    Last change on 5th of November 2025, at 9:46 (CET) | Publication Standard 1.0
    Publication category
    Impactful algorithms
    Impact assessment
    DPIA
    Status
    In development
  • At the civil affairs counter, old documents submitted for replacement for a new, valid one must be scanned for authenticity. The document scanner scans the authenticity characteristics of a driving licence, identity card or passport. If authenticity is established, the resident will receive a new document.
    Last change on 23rd of January 2025, at 15:52 (CET) | Publication Standard 1.0
    Publication category
    Impactful algorithms
    Impact assessment
    DPIA
    Status
    In use
  • At the civil affairs counter, old documents handed in for replacement for a new, valid one must be scanned for authenticity. The document scanner scans the authenticity characteristics of a driving licence, identity card or passport. If authenticity is established, the resident will receive a new document.
    Last change on 19th of October 2025, at 19:05 (CET) | Publication Standard 1.0
    Publication category
    Impactful algorithms
    Impact assessment
    Field not filled in.
    Status
    In use
  • This algorithm helps employees review documents. They can search, group documents and find duplicates. This allows reviewers to quickly see whether parts of documents are important and whether the information may be public.
    Last change on 4th of June 2026, at 14:00 (CET) | Publication Standard 1.0
    Publication category
    Other algorithms
    Impact assessment
    Field not filled in.
    Status
    In use