Hungarian Online Job Advertisements – Experimental Statistics

ongoing

Building upon the previous work and experience on the Web Intelligence Challenge (classification of online job advertisements) the HCSO’s Jobcatchers team – taking up with new members – wishes to continue innovating and producing new kind of experimental statistics using OJA data. While our main goal is to produce the aforementioned experimental statistics that could be regularly updated, along the way we can gain various benefits. For example developing new AI solutions, fine-tuning ML algorithms, utilising web scraping, establishing collaborations with data holders. Since this is a complex task, we extend our team with labour market, IT, dissemination and coordinator experts to make the product as relevant and comprehensive as possible. We want to identify potential sources, transfer data from Hungarian online job portals, scrape the carrier sites of the companies with the largest number of employees in Hungary and also use WIH data. After the database is set up we plan to further develop our ISCO coding solutions. Based on a thorough analysis and quality measures we intend to present the findings in the most useful form for the users (e.g. most in-demand skills, tendencies in different ISCO categories, distributions in time and location), moreover we also intend to return some advantageous result to the data providers (e.g. keywords regarding ISCO codes). From a methodological and IT point of view we plan to use open source tools as before, but train more sophisticated neural networks on a bigger scale.