Periodic Reporting for period 1 - FEAST (Fair, Effective, and Sustainable Talent Management using Conditional Network Embedding)
Reporting period: 2021-04-01 to 2022-09-30
Public and commercial employment agencies as well as commercial Human Resources companies are interested in making the job market more efficient, by ensuring the best match between jobs and employees, giving workers optimal career prospects and job satisfaction, and maximizing productivity. Increasingly, they make use of Artificial Intelligence (AI) methods to accomplish these goals.
The FEAST project investigated the potential of Conditional Network Embedding, a platform AI technology that allows creating a joint representation of all relevant entities in the job market: jobs, job seekers, work and home addresses, skills and competences, training courses, and more. Built upon it, specific services can be developed with limited effort, such as job matching, identifying skills gaps, giving curriculum advice, identifying new and emerging skills in the job market, and HR strategy and recruitment tools.
Specific to the FEAST approach is that the joint representations of all relevant entities can be debiased. Any biases present in the underlying data can be removed from the data, such that the downstream tasks are debiased as well.
In this project, in consultation with industry advisors, we investigated the market potential of the platform and these use cases, strengthened the IPR position, and made important technical advances on the FEAST platform.
The FEAST project investigated the potential of Conditional Network Embedding, a platform AI technology that allows creating a joint representation of all relevant entities in the job market: jobs, job seekers, work and home addresses, skills and competences, training courses, and more. Built upon it, specific services can be developed with limited effort, such as job matching, identifying skills gaps, giving curriculum advice, identifying new and emerging skills in the job market, and HR strategy and recruitment tools.
Specific to the FEAST approach is that the joint representations of all relevant entities can be debiased. Any biases present in the underlying data can be removed from the data, such that the downstream tasks are debiased as well.
In this project, in consultation with industry advisors, we investigated the market potential of the platform and these use cases, strengthened the IPR position, and made important technical advances on the FEAST platform.