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.
Information on Supported Treatment – Short-Stay (Schengen) Visa (KVV)
- Publication category
- Impactful algorithms
- Impact assessment
- DPIA, IAMA
- Status
- In use
General information
Theme
- Organisation and business operations
- International
Begin date
Contact information
Link to publication website
Link to source registration
Responsible use
Goal and impact
The Ministry of Foreign Affairs (BZ) is responsible for processing and assessing applications for short-stay visas (KVV) for the Schengen countries. In doing so, the BZ fulfils the role relating to (regular) admission within the migration chain.
Before a KVV can be issued, the BZ must assess whether all the mandatory conditions under European legislation and regulations have been met. Due to the large volume of applications, but also to carry out this task as objectively and carefully as possible, the Ministry of Foreign Affairs supports this process with an algorithm designed to facilitate a data-analysis-based approach. This is known as Information-Supported Processing (IOB).
The IOB is used to estimate the expected intensity of the processing of a visa application. The estimate of the expected intensity refers to the expected use of resources. For example, conducting an additional interview or requesting further necessary information. This assessment is presented to the consular officer in the form of a so-called ‘track’ (fast, regular or intensive), accompanied by additional relevant information.
The conditions for obtaining a visa are and remain the same for everyone, regardless of previous applications or those for which a sponsor has provided a guarantee. These conditions have been agreed between Schengen countries and are laid down in the EU Visa Code. The use of the algorithm supports the processing of a visa application but is never a ground for refusing a visa. The impact of using the algorithm is therefore minimal.
Considerations
It is not possible for consular staff to manually check all relevant information at the time of application. Thanks to the algorithm’s recommended track, they can estimate the expected workload involved in processing a specific visa application and the associated file review. This enables them to better assess whether, for example, additional information is required or whether an extra interview needs to be conducted.
Furthermore, the processing of applications becomes more objective, as it is no longer based solely on the consular officer’s judgement, but is also supported by data. This creates a better balance between the available information and the consular officer’s knowledge and experience.
Human intervention
It is, and remains, up to the consular officer to decide what the next step in the processing of the application should be and whether a visa will ultimately be granted or refused. The track is therefore merely a piece of supporting advice to the consular officer regarding the processing of the application and never concerns the decision to be taken on the application itself.
The recommended track is expressly not a ground for refusal. After all, when processing the application, the consular officer has the entire file at their disposal, as well as, of course, their previous experience and knowledge. This therefore constitutes (significant) human intervention.
Risk management
Transparency and explainability were decisive factors in the choice of algorithm. The decision tree’s operation is clear. Each step can be followed and explained. Furthermore, the rules are laid down in advance and do not change unless they are manually adjusted by BZ itself. This makes the algorithm verifiable: the decision paths generated by the algorithm are reproducible.
BZ monitors the final decisions on an application. These decisions are compared with the track generated by the algorithm. This ensures that decision-makers are not simply ‘following the track blindly’. In addition, a manual check is carried out whenever there is a match with a source.
Legal basis
Performance of a public task (Article 6(1)(e) of the GDPR, Article 21 of the Visa Code, Article 2(2) of the National Act)
Links to legal bases
- National Visa Act: https://wetten.overheid.nl/BWBR0038494/2018-07-01
- Visa code: https://eur-lex.europa.eu/legal-content/NL/TXT/HTML/?uri=CELEX:32009R0810
- GDPR: https://avgb.nl/art-6-avg/
Elaboration on impact assessments
A summary has been drawn up for the IAMA. This is included as an annex to the IAMA.
Impact assessment
- Data Protection Impact Assessment (DPIA)
- Human Rights and Algorithms Impact Assessment (IAMA): https://open.overheid.nl/details/cf6901e1-0b55-418d-91cc-5e388f83ebb0
Operations
Data
The visa application is assessed on the basis of the following information: (1) information submitted to the Ministry of Foreign Affairs by the applicant themselves; (2) information from the sponsor and/or employer; (3) information from the migration chain; and (4) information from profiles based on similar applications from the past.
For a complete overview of the data, see also the Factsheet on Information-supported Processing.
Technical design
The algorithm can be broken down into three methods used; not every method can be classified as an algorithm:
1. Hit/no hit search in the available data sources (not an algorithm);
2. Decision tree for the purpose of creating profiles (algorithm);
3. A weighting model for determining a score and assigning a track (algorithm).
The ‘hit/no hit’ method is merely a comparison method and not an ‘algorithm’. It simply checks whether there is a match between the data from a visa application and the data sources held by the Ministry of Foreign Affairs.
The decision tree consists of a series of pre-defined ‘if–then’ rules. This is a simple type of algorithm that operates on fixed rules: if a certain condition is met, then a specific step or outcome follows. The rules are drawn up in advance by the Ministry of Foreign Affairs itself. The algorithm therefore does not learn or change on its own, but merely applies these fixed rules.
The weighting model is a computational model used to assign a score, and ultimately a track. The weighting model can be classified as an algorithm.
Further information on how it works technically can be found on GitHub: https://github.com/HDCV-Data/bao-broncode.