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.

Signals

public space notifications

Last change on 23rd of July 2024, at 6:11 (CET) | Publication Standard 1.0
Publication category
Other algorithms
Impact assessment
Field not filled in.
Status
In use

General information

Theme

Space and Infrastructure

Begin date

01-2023

Contact information

gemeente@zundert.nl

Link to publication website

Melding maken - gemeente Zundert

Responsible use

Goal and impact

If something needs to be made or cleaned up on the street or in a park, this can be reported to the municipality via SIA, the online reporting system. A dangerous traffic situation or nuisance caused by people and catering establishments can also be reported. Previously, people had to choose which category their report best fit (for example 'nuisance' or 'street furniture'), so that the report would be sent to the correct department of the municipality. However, the municipality is a complex organisation and the list of categories is long. As a result, the correct category was not always chosen. This sometimes caused delays in the processing of reports. That is why we now use an algorithm that recognises words, such as 'waste' and 'pavement'. Based on this, it is determined which category the report best fits and which department should process the report. The reporter no longer has to choose a category, and the report is processed more quickly because it is sent to the correct department.

Considerations

supplier independent, common ground principle

Human intervention

All reports that are classified with less than 40% certainty in a certain category are forwarded to the KCC. A human assessment then takes place and the report is still categorized. Reports that are wrongly forwarded to an incorrect category are also manually placed in the correct category by the responsible department (sometimes via the KCC).

Risk management

There are few risks associated with this algorithm. It places a report in the correct category and ensures that it is brought to the attention of the correct department more quickly. If the algorithm cannot place a report in a category with sufficient certainty, it ends up in the 'Other' category. KCC employees review these reports and manually place them in the correct category. If the algorithm does not work properly, it will take a little longer for the report to reach the correct department. The reporter can add personal data if he or she wants to be kept informed. This data is stored securely and is not used by the algorithm.

Operations

Data

notifications


The dataset consists of reports that have been made previously (free text field). Initially, we used 30,000 reports from the past years to train the algorithm. It is regularly retrained with new reports and corrections made to existing reports. If the KCC or the service department sees an incorrect categorization (see Human supervision), they correct this manually in the reporting system. These corrections are used for retraining. We are investigating whether the retraining of the algorithm can be automated in the future.


We cannot make this dataset public in this register. Because the data comes from a free text field, it may contain personal data, although this is not explicitly requested.


Email address and telephone number for follow-up questions


Someone who reports can leave their phone number and/or email address if they want to. We will then keep the reporter informed of the progress and we have the option to call them back. This information is not stored longer than necessary for this purpose and is therefore not used by the algorithm.


Technical design

Architecture of the model

The text of the report is divided into separate words. Each word from a report is analyzed to see how unique it is for that report, compared to the total collection of words ('TF-IDF' or 'term frequency-inverse document frequency'). A word like 'de' or 'bedankt' is given a low weight and a word like 'garbage' is given a higher weight.

Logistic regression (a machine learning technique) is then used to determine which category the report belongs to and thus which department within the municipality the report most likely belongs to.

External provider

VNG