Next career recommendation in Mississippi with artificial intelligence

Abstract

Connecting unemployed people to job openings has been a challenge post-pandemic. With the help of artificial intelligence and big data, we addressed this issue by creating a deep learning model to provide realistic job recommendations for unemployed people based on the employment history of each individual. First, the transfer learning model was applied to match job titles and O*NET Standard Occupational Classification (OSOC) codes using data on job seekers from the Mississippi Department of Employment Security, where OSOC is a standard occupational classification-based system used by U.S. federal agencies to classify workers into occupational categories. Next, a Long Short-Term Memory (LSTM) model was created for career pathway prediction, to generate the top three OSOC job recommendations based on the individual’s employment history. The final model accuracy was 72.8% when an individual’s education history was included.