Input help Business activities 2.0

Chamber of Commerce

Supporting start-up entrepreneurs in drafting the (business) activity description, during registration in the Trade Register by means of question and answer sessions

Last change on 18th of November 2024, at 9:49 (CET) | Publication Standard 1.0

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

General information

Theme

Organisation and business operations

Begin date

2024-11

Contact information

[e-mail]

Link to publication website

https://www.kvk.nl/inschrijven/inschrijven-en-afspraak-maken-eenmanszaak/

Responsible use

Goal and impact

The goals are:

1. The registration of a sole proprietorship for start-ups has been extended with the Input Help. This allows the start-up to independently prepare a unified, structured activity description through question-and-answer with artificial intelligence.

2. Simplify the translation of business activities into SBI codes for the KVK employee. Because the supplied texts are of higher quality and suggestions for appropriate SBI codes are made, classification can be faster and easier and the registration interview can be about starting the business.

Considerations

Actors:

• (Aspiring) Entrepreneurs
• Chamber of Commerce employees

Interests:

The interests and frameworks of the two groups are laid down in the Trade Register Act. This application does not affect the interests of the KVK employees, other than improving data quality.

For the (prospective) entrepreneurs, their (individual) interests are better served by this solution as he/she is better assisted in purchasing mandatory services

Values:

The solution meets the values fairness, freedom. The value equity does not directly apply, as large volume groups are served first. This does not mean that groups are excluded, but that common situations are served even better. This is similar to the top sector policy and/or top task policy.

Human intervention

• Entrepreneur can overwrite/amend suggestions in pre-registration
• Entrepreneur can change the description during (mandatory) registration/registration at the front office (at a Chamber of Commerce office)
• Entrepreneur can change it after registration/registration

There is no automated decision.

Risk management

• There is no automated decision
• There is no processing of personal data
• The process is monitored for Customer Satisfaction (CSAT) and Customer Effort Score (CES)

Legal basis

Trade Register Act section 13, Trade Register Act 2007

Links to legal bases

Handelsregisterwet: https://wetten.overheid.nl/jci1.3:c:BWBR0021777&hoofdstuk=2&paragraaf=2.2&artikel=13&z=2024-06-19&g=2024-06-19

Elaboration on impact assessments

• Prescan of DPIA done, which showed that no DPIA is required
• BIA - Business Impact Analysis has been carried out

Impact assessment

• Prescan DPIA
• BIA

Operations

Data

Using entered text, follow-up questions help the entrepreneur to arrive at a correct (business) activity description. Ideally, this leads to a single SBI code.

Links to data sources

• CBS SBI codelijst : https://sbi.cbs.nl/
• NLTK: https://www.nltk.org/
• Spacy: https://www.spacy.io
• Spellingchecker: https://www.opentaal.org/
• Handelsregister: https://www.kvk.nl/zoeken/

Technical design

1. Entrepreneur completes activity description
2. A spell check is performed
3. Input help checks whether there are multiple activities in the description
4. Activity description is classified and shows 3 most appropriate SBI code (with certain certainty) for the branches construction, web shops and health care
5. For the other branches, the entrepreneur gets a short description of the proposed/classified SBI code (summary of the data from CBS).
6. This is followed by a question to check whether this is correct. If yes: entrepreneur continues with own activity description
7. Entrepreneur is still free to modify the proposed activity description.

External provider

Internally Developed

Link to code base

https://rasa.com/docs/rasa/components#dietclassifier-1 https://rasa.com/docs/rasa/policies#ted-policy https://lightgbm.readthedocs.io/en/latest/pythonapi/lightgbm.LGBMClassifier.html https://github.com/mwydmuch/extremeText