The Employment Pilot Project aims to build a user-friendly algorithmic matching system that works to smoothen job recruiting and job seeking processes.
The Covid-19 pandemic is a black swan event that has brought major changes into the employment market worldwide, from enabling and normalizing remote work to great resignation where many people have changed the direction of their careers or contributed to a major increase of self-employment.
The employment market in Belgium is no exception. According to Statbel (the Belgian statistical office), there was an increase in the number of both employed and unemployed people in Belgium from 2020 to 2021, as the result of the Covid-19 crisis. In 2021, an average of 4,854,000 people living in Belgium were employed which was an increase of 1.1% (51,000 people). And at the same time the ILO unemployment rate in Belgium also rose to more than 14.8% (42,000 people) and those people are people who do not have a job and are actively seeking for it, and they are able to start working within 14 days.
FARI has introduced its first pilot project on employment since October 2021 with a close collaboration with Actiris, Brussels’ official employment platform that aims to provide employment solutions to both job seekers and employers in the region. This Employment Pilot Project aims to build a user-friendly algorithmic matching system that works to smoothen job recruiting and job seeking process by training Artificial Intelligence to automate the formatting of job offers and job seekers’ profiles in the existing system of Actiris. It also aims to ensure a better matching process for both job stakeholders.
Amid the process of development, the pilot project encountered some existing issues including the institutional constraints that required the project to use a formatted catalog of jobs, for example ROME 3 link system in French (also used in Wallonia) and the catalog used in Flanders in Dutch. This catalog tends to be biased since it restricts the possibilities of the matching for emerging jobs and for under-represented sectors. This happens because the matching relies entirely on the Pure Algorithmic approach that searches only for near perfect matches, leading to empty list results and it leaves no room for Free-Form Text Matching, an effective approach to avoid the format step needed by the algorithms. The idea here is to relax constraints and make room for compromises between the perfect candidate that recruiters look for and the actual job seekers.
What makes this Employment Pilot Project unique is the fact that for Brussels, the project needs to deal with three languages at the same time (French, Dutch, and English). The catalog plays an important role in mapping between texts and the references of the job or skills and hence is independent from the language while other solutions adopted by projects in other countries only focuses on one language, for example: English in the UK and Japanese in Japan.
Additionally, most employment agencies benefit from research on the subject that only deal with highly skilled profiles and only prioritise the need from the employers’ side. This is why FARI’s Employment Pilot Project prioritises helping mostly low profile job seekers who really need a job and are open to do anything that do not require a degree in a specific field. There is also other research related to this use-case conducted in France and Flanders with a need to focus on putting people to work more than satisfying the recruiters.
What is also needed here is a regular updating on the job catalog of the Actiris platform, especially that more skills are getting in demand now. For example: IoT is a new in demand skill that did not even exist five years ago, so it needs to be added into the catalog. The pilot project has another bigger aim than enhancing a matching system for Actiris, it also hopes to accelerate itself to building on a Recommendation System that is able to make use of search history data of the users (e.g clicks) to offer smart suggestions on possible job opportunities for job seekers and not just trapping them within the bubble of the texts that highlight their skills and experience but open new doors for those who are looking for a shift in their current career or expertise. Such recommendation algorithms are known to be used by Netflix, YouTube, Amazon, etc.
The FARI team has been working closely with the IT team of Actiris that have helped identify the actual needs of the public administration and outline the solution to implement together. The metric of the success of this project will be determined by how well the solution integrates into the Actiris environment. Both teams are working together to integrate the developed solution into the current Acritis software.