Information extraction from, and classification of, notarial deeds (DNI)

Dutch Tax and Customs Administration

The algorithm detects and selects text from notarial deeds for monitoring the correct registration of the notarial deed and levying and collecting taxes.

Last change on 25th of June 2024, at 10:22 (CET) | Publication Standard 1.0

• Publication category: Impactful algorithms
• Impact assessment: DPIA
• Status: In use

General information

Theme

Public finance

Begin date

Field not filled in.

Contact information

[e-mail]

Link to publication website

https://over-ons.belastingdienst.nl/onderwerpen/omgaan-met-gegevens/algoritmeregister/

Link to source registration

https://over-ons.belastingdienst.nl/onderwerpen/omgaan-met-gegevens/algoritmeregister/informatie-extractie-uit-en-classificatie-van-notariele-akten-dni/

Responsible use

Goal and impact

The Tax Authority uses information from notarial deeds for the purpose of supervising the registration of notarial deeds, in addition the data is used for levying and collecting taxes such as inheritance, gift and transfer taxes.

Considerations

Large numbers of deeds are involved, which is why we extract the information from the deed largely automatically using the algorithm.

By using the algorithm, fewer people are needed to manually search the information in the deeds and the information from a deed becomes available digitally faster. Deeds can therefore be processed faster. In addition, the use of the algorithm ensures that the deed is available in the right place within the organisation.

Human intervention

If a deed type cannot be determined, or if other relevant information is missing from the deed, the notary is contacted - if manual research by the Inland Revenue does not provide a solution - to complete or correct the missing information.

Risk management

1. The use of the data has been tested against the General Data Protection Regulation (AVG) through a Data Protection Impact Assessment (DPIA). The Inland Revenue prevents direct discrimination with algorithms. Special personal data, such as ethnic origin, do not play a role.
2. Both algorithms were developed in accordance with the Tax and Customs Administration's quality framework. This contains rules and agreements that were followed during algorithm development. The conditions of the National Audit Office are leading in this respect. At set moments, the Tax and Customs Administration checks whether the algorithm still meets the quality requirements.
3. The algorithm was developed at the Tax and Customs Administration itself and is also maintained internally. By arrangement, the team that developed the algorithm and the team that performs functional management regularly check whether the results are of sufficient quality.

Legal basis

- Registration Act 1970

- Regulation implementing the Registration Act 1970

- 2003 Tax Administration Implementation Regulations

- Notaries Act

- General State Taxes Act

- General provisions Citizen Service Number Act

- Archives Act 1995

Links to legal bases

• Registratiewet 1970: https://wetten.overheid.nl/BWBR0002739/
• Uitvoeringsregeling Registratiewet 1970: https://wetten.overheid.nl/BWBR0034017
• Uitvoeringsregeling Belastingdienst 2003: https://wetten.overheid.nl/BWBR0014506
• Wet op het notarisambt: https://wetten.overheid.nl/BWBR0010388
• Algemene wet inzake rijksbelastingen: https://wetten.overheid.nl/BWBR0002320/
• Wet algemene bepalingen Burgerservicenummer : https://wetten.overheid.nl/BWBR0022428/
• Archiefwet 1995: https://wetten.overheid.nl/BWBR0007376/

Impact assessment

Data Protection Impact Assessment (DPIA)

Operations

Data

• PDF of notarised deeds
• Name
• Address
• Residence
• Date of birth
• Place of birth
• Chamber of Commerce number
• Special personal data (Pursuant to sections 7a and 7b of the Registration Act 1970, the Tax Authorities receive and store full copies of notarial deeds. Inherently, data that qualify as special personal data can be distilled from the text of the deeds, such as directors of organisations from which a political preference, ethnic origin or religious opinion can be traced. Special personal data are an integral part of receiving and storing deeds, but play no role in further processing.

Technical design

Two algorithms are used in one logical context.

1. Algorithm 1 determines what kind of deed has been received
2. Algorithm 2 extracts relevant information from deed.

Algorithm 1 uses "machine learning". The algorithm is not self-learning, which would mean that it evolves as it is used. It doesn't.

Algorithm 2 includes a set of rules that extracts information from deeds, for example names, addresses and dates of birth. For specific deed types, roles are also extracted from the deed (who is the buyer, testator, notary, etc.).

This information is then available to the departments that need the information to carry out their work, namely:

• The Royal Notarial Association (KNB) and the notary himself for the purpose of monitoring the correct registration of the notarial deed; and,
• The Inland Revenue directorates: Individuals, Small and Medium Enterprises and Large Enterprises for the purpose of levying and collecting taxes.

Both Algorithm 1 and Algorithm 2 are in support of the overarching process of 'Digitisation of notarial information'.