Case Matcher: A search and find function in asylum applications

IND

Staff processing asylum applications can use this algorithm to search for similar applications to treat applications equally.

Last change on 21st of May 2024, at 10:05 (CET) | Publication Standard 1.0

• Publication category: Other algorithms
• Impact assessment: Field not filled in.
• Status: In use

General information

Theme

Migration and Integration

Begin date

07-2021

Contact information

[e-mail]

Responsible use

Goal and impact

Asylum applications should be treated fairly and equally. This algorithm allows the clerk to search and find all similar asylum applications faster. For example, the clerk searches by year and document type such as application, decision, ruling and client information. Asylum applications that are very similar can thus be treated similarly. The caseworker can also find information needed to process the application faster.

Considerations

Asylum applications can be processed faster, more fairly and equally if the caseworker can search and find similar asylum applications. This algorithm allows the clerk to search and find more easily, instead of going through previously processed applications piece by piece. Using search terms, a list of documents and asylum applications that have similarities with the search terms appears. The employee chooses from the list and opens the document and/or application to go through the information. The clerk can choose to copy, for example, the legal grounds on which an application has been approved or rejected in the past. This may allow the employee to process the application faster.

Human intervention

The caseworker searches for words and gets a list of all documents and asylum applications that contain these words. The caseworker reviews the list and assesses whether it contains relevant information to process the application. Due to changes in policy, a previous decision on an asylum application may no longer fit the current policy. The caseworker is obliged to decide based on the policy currently in force. Incidentally, each new asylum application is always read and assessed by several experts.

Risk management

Only employees with permission have access to this algorithm. Employees must also undergo training before they can use this algorithm; this is part of the internal IND training. This training explains how the employee can and may use this search and find function. For example, the employee is not allowed to search on persons (names). The IND is bound by privacy rules under the AVG. This means that documents and applications may not be viewed just like that. The employee himself is always the expert and the decision-maker (and not the algorithm). The employee may not judge solely on the results of the algorithm, but must always take into account the current policy, applicable instructions and current events.

With experts, which documents are relevant in an asylum application are assessed in advance. This ensures that only relevant documents are found in the search.

Legal basis

Aliens Act 2000, sections 107 and 107a

Operations

Data

The algorithm searches a database of asylum applications and relevant documents such as appeals. For example, the clerk can search by: case number, document type, country of origin, nationality, gender or year.

Technical design

The algorithm determines the relevance of documents and asylum applications eligible to be searched and found through that finding system. The relevance is determined in 2 steps:

1. The employee can filter in the main screen by, for example, year and nationality (which is a highly relevant variable in an asylum application) and can search by terms. These search terms are then searched in the filtered documents and the documents where they occur (whether in combination or not) come up in a list. This list of documents is ordered by relevance; the higher, the more (possibly) relevant these documents are. This relevance score is determined by standard search algorithms. A document is ranked higher if terms occur more often and also in combination with each other. The employee can then check the documents themselves to see if they are relevant to the case to be handled.

2. The employee can also choose to have the found documents aggregated (merged) at case level. The documents that were filtered and in which the search terms came up are now listed per asylum application and presented in a table. Now the employee does not see a list of documents but a list of asylum applications, and again, the higher the asylum application is in the list, the more relevant it may be. In this case, relevance is determined by a scoring mechanism based on document type.

With the help of experts, it has been determined that some documents are more important than others. Consider an asylum narrative (which describes the story of the asylum seeker) or a document summarising the asylum application. In the overview of asylum applications, the employee can quickly see in which documents the search terms occurred.

External provider

Knowledge Plaza