Organisation Name

Nobl.ai

Industry

Postal and telecommunications services

Organisation Website

https://nobl.ai/

Country

Belgium

Sustainable Development Goals (SDGs)

General Description of the AI tool

The Nobl.ai Job Matching Engine is an AI-driven tool designed to find the right candidate for every job vacancy and the right vacancy for every candidate. The system is built with trustworthiness and explainability as core values. It also features bias detection and mitigation strategies, ensuring equal opportunities for all users, regardless of their demographic or socio-economic backgrounds. As data quality is often a challenge in developing job markets, the system can provide recommendations purely based on user behaviour or a combination of user data and user behaviour.

Relevant Research and Publications

The research papers summarized below have one or more authors affiliated with Nobl.ai and have inspired and shaped the technology developed by the company.

"Conditional Network Embeddings", Kang et. al, 2019. Sets the foundations of a probabilistic model for creating vector representations that are independent of specific data properties.
"DeBayes: a Bayesian Method for Debiasing Network Embeddings", Buyl et. al., 2020. Extends the previous work to computing vector representations that are independent of properties such as age, gender, race, etc. Building on this work, we can ensure that the recommendations our model provides are bias-free.
"SkillGPT: a RESTful API service for skill extraction and standardization using a Large Language Model", Li et. al., 2023. Our AI system utilizes skills to better match users and jobs. This work explores ways of extracting and standardizing these skills.
"ReCon: Reducing congestion in job recommendation using optimal transport", Maskayekhi et. al., 2023. This work addresses the critical challenge of congestion when too many job seekers and too few jobs (or vice versa) are available in a specific subsector of the labour market. It provides tools to alleviate these situations, which we borrow for our AI system to imporve the overall performance of the labour market.
"A challenge-based survey of e-recruitment recommendation systems", Maskayekhi et. al., 2024. Provides an extensive analysis of the e-recruitment landscape and the main challenges models face, as well as how these are tackled in academic settings. These challenges defined and inspired the main design principles behind our AI engine.

Needs