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.

Algolia AI Search - Retrieving search results website(s)

This algorithm allows users to quickly and easily search the website(s) using Algolia. It provides features such as a search field, filters, infinite scrolling, search suggestions and sorting of results. Thus, users easily find what they are looking for and discover new products and information.

Last change on 3rd of June 2025, at 11:23 (CET) | Publication Standard 1.0
Publication category
Other algorithms
Impact assessment
DPIA
Status
In use

General information

Theme

Organisation and business operations

Begin date

04-2025

Contact information

algoritmes@arnhem.nl

Link to publication website

opendata.arnhem.nl

Responsible use

Goal and impact

The purpose of this algorithm is to allow users to quickly and easily search Arnhem.nl and its sub-sites. Algolia is a cloud service that is used for this purpose and searches an index that the municipality of Arnhem itself can set up and adapt to its needs. Citizens and businesses interact with the algorithm when searching for information, products or services on the website; they are then quickly presented with relevant and well-organised search results.

Considerations

Using Algolia as the search algorithm for Arnhem.nl offers many advantages. Users find relevant information faster and easier, even if they make typos or enter incomplete search queries. Ethically, it is important that search results are fair and transparent and that user privacy is respected. Algolia offers options to customise the search index to exclude unwanted or sensitive information. The choice of Algolia is justified because it greatly improves the search experience for citizens and businesses, speeds up services and future-proofs the website.

Human intervention

Using Algolia for search on Arnhem.nl usually does not require human intervention in displaying search results. Users are automatically shown results instantly based on their search query. However, employees of the municipality can control and adjust the algorithm and the search index, for example by adding filters, sorting results or excluding unwanted information. Thus, they can adjust the algorithm's operation if necessary. So there is no direct control on every single search result, but the results can be periodically checked and improved by humans. This keeps the search function reliable and allows errors or unwanted results to be corrected.

Risk management

When using Algolia as a search algorithm, there are several risks, such as technical risks, errors in search results and dependence on a third-party vendor.

To mitigate these risks, several measures are taken. The system is regularly tested and monitored so that failures and errors are quickly noticed and resolved. The operation of the algorithm is checked for compliance with laws and regulations. Also, the search index is designed to exclude unwanted or sensitive information and care is taken to ensure fair and neutral search results to avoid discrimination. Users can provide feedback and adjustments are made where necessary.

In addition, Algolia meets high international security standards. Algolia is SOC 2 and SOC 3 compliant, ISO-certified and offers encryption for additional data protection. This approach minimises risks and keeps the search algorithm reliable and responsible in use.

Legal basis

When adding a search engine to a government website, Article 3 of the Government Digital Accessibility Decree is particularly relevant. The legal basis for this is the Digital Government Act. This means that the search functionality must comply with the accessibility requirements laid down

Links to legal bases

  • Wet digitale overheid: https://wetten.overheid.nl/BWBR0048156/2023-07-01
  • Besluit digitale toegankelijkheid overheid, artikel 3: https://wetten.overheid.nl/BWBR0037987/2024-04-01

Elaboration on impact assessments

Pre-DPIA conducted. Full DPIA not necessary

Impact assessment

Pre-DPIA

Operations

Data

Our search engine processes only the search terms entered by users to display relevant information on the website.


  • No personal data is processed, stored or used.
  • No information from basic registries (such as BRP, BKR) or organisation-specific customer data is used.
  • The search terms entered are not stored and not used for training AI models.
  • Our data will never be stored or used without permission for purposes other than showing search results.


In short, no personal data or other traceable data will be used or stored. The search engine works exclusively on the basis of anonymous searches and does not process any privacy-sensitive information.

Technical design

Technical operation:

The search engine uses the Algolia platform, which employs a non-self-learning, deterministic algorithm to rank search results.

Input:

Users enter a search term or query via the search field on the website.

Operation:

  • Algolia pre-indexes all relevant web pages and content of the website. This is done by crawling the content and structuring it into a searchable index.
  • When performing a search, the algorithm compares the entered search term with the indexed data.
  • Algolia uses a tie-breaking algorithm: when multiple rules apply to a query, the algorithm determines the order of search results based on preset criteria. More specific rules are usually given higher priority.
  • The algorithm uses multiple factors such as textual relevance, field priorities and any set filters, but no self-learning optimisation takes place.
  • No personal data is used or stored in the process.

Output:

The search engine displays a sorted list of relevant results based on the query. The output consists only of website content matching the search terms.

Characteristics:

  • Not self-learning: the algorithm does not automatically adapt based on user behaviour or new data.
  • No use of neural networks or complex machine learning models.
  • Fully deterministic and transparent in its operation.


External provider

Algolia Inc.

Similar algorithm descriptions

  • We want to use AI avatar videos to target groups of citizens we find difficult to reach (non-native speakers, low-literate, people who have difficulty accessing accurate & reliable information) and improve access to accurate and important municipal information, facilities, and programmes (+acts).

    Last change on 5th of February 2025, at 12:14 (CET) | Publication Standard 1.0
    Publication category
    Impactful algorithms
    Impact assessment
    IAMA, ...
    Status
    In use
  • The search algorithm used by the BSN management facility to search for individuals based on identifying data.

    Last change on 25th of September 2024, at 13:20 (CET) | Publication Standard 1.0
    Publication category
    Other algorithms
    Impact assessment
    DPIA
    Status
    In use
  • This algorithm helps retrieve information in cold case files. It uses a language model to search for the meaning of words and not just the exact words.

    Last change on 28th of January 2025, at 13:28 (CET) | Publication Standard 1.0
    Publication category
    Other algorithms
    Impact assessment
    DPIA, ...
    Status
    In use
  • Algorithm that enables phonetic (writing data as they sound) searches on personal data of foreigners registered in the Basisvoorziening Vreemdelingen (BVV).

    Last change on 21st of December 2023, at 15:38 (CET) | Publication Standard 1.0
    Publication category
    Other algorithms
    Impact assessment
    Field not filled in.
    Status
    In use