Text analysis for job advertisement (Swedish Unemployment Agency)

Potential supervisors: 

This project concerns text analysis and natural language processing for job ads, and will be conducted in collaboration with the Swedish Unemployment Agency. In particular, the students will work on extracting structured meta-data from unstructured natural language text from job advertisments. This meta-data can then be used for job searches by individuals. Extracting reliable meta-data from natural language is a challenging problem involving e.g. NLP-techniques as well as machine learning.

Arbetsförmedlingen (The Swedish unemployment agency) is collaborating with private owned Job-Boards and Statistics Sweden (SCB) in order to get better statistics of the labour market and better user search experience. The challenge is to be able to offer analytics based on all ads regardless of where they are published or in which format and standard. An important use case is to help job seekers that have been out of a job for a longer period of time. Arbetsförmedlingen daily process around 75 000 ads and need to present those that are most relevant for this segment. Ads are unstructured data. We need help using text analytics to produce metadata from unstructured text, for the necessary search and filtering parameters. With NLP and machine learning  we think it possible to collect data from different sources and structure it for a better job seeker experience. Data for the project will be provided by Arbetsfömedlingen.

For more information you may contact Maria Dalhage at Arbetsförmedlingen's departement Jobtechdev: maria.dalhage@arbetsformedlingen.se

Date range: 
August, 2021 to August, 2026