In many countries, anyone wishing to apply for a passport must not only expect a processing time of several weeks, but also several visits to government offices. With Veridos’ identity solutions, this process can be reduced to a matter of days or even hours. A key component of this is the digital application process via an online platform: citizens submit their details and documents online, book an appointment and thus save valuable time.
The challenge: The application data entered often differs from the existing databases, meaning that staff at the relevant authorities have to manually correct the data, which is very time-consuming. This is where iteratec’s expertise in machine learning (ML) comes into play.
Through the proof of concept, our team of software developers and machine learning specialists has demonstrated that the manual effort involved in validating application data can be significantly reduced. By using ML algorithms, the number of automatically validated applications can be significantly increased, thereby speeding up the passport application process.
Another key requirement of the project was the transparency of the decisions made by the ML model when verifying application data. These decisions are made transparent within the system and presented in a way that is clear to staff.
When an applicant enters their details on the website to apply for a passport, these are cross-checked against data held by various authorities. Discrepancies frequently arise during this process, necessitating manual follow-up work. The requirements set for us were:
In the first step, an analysis of the data was to identify the most common causes of manual follow-up work, in order to evaluate effective approaches for increasing the number of cases processed automatically.
Based on the decisions previously made by case officers, a machine learning model was developed to predict the outcomes of manual checks. The model serves as the basis for further automation of the decision-making process.
The country-specific circumstances posed a challenge: names can be transliterated into the Roman alphabet in various ways. This leads to inconsistencies, which the model must either accept or reject accordingly.
It was particularly important to rule out false positives – that is, applications containing invalid data – as these would necessitate rework due to invalid data at the latest when the passport was issued.
None of the methods used to date were able to provide satisfactory support for the automated verification of data. Our proof of concept demonstrated how the use of ML can significantly increase the number of automatically validated applications. In the prototype we developed, the decisions made by the ML models were also presented to staff in a transparent and comprehensible manner.
Both the agile project approach and the prototype developed form the basis for the expansion and productive deployment of the system.
Manual effort can be significantly reduced.
The time taken to apply for passports can be shortened.
We support the digitalisation of public administration.
The collaboration with iteratec on this proof of concept was excellent. The team quickly familiarised themselves with the subject matter and were able to provide valuable input within a very short time. The agile working method enabled rapid results, which were continuously improved. We look forward to the next steps!
Senior Product Manager, Veridos GmbH
As one of the leading providers in its field, Veridos offers modern solutions for border controls and for the production and issuance of passports, ID cards, health insurance cards, driving licences and vehicle registration certificates. The company was founded in 2015 as a joint venture between Giesecke+Devrient and Bundesdruckerei. With around 600 employees, Veridos has a global presence across various locations.
