Job profile demand understanding in international financial organizations: a natural language processing approach

Abstract

The present work aims to create a Machine Learning model using unstructured text data in order to predict whether a position is prone to taking a longer Time to Fill than the overall average. As well as building an initial categorization of profiles within the organizations in which it was carried out, this will help to provide insights that allow understanding both the demand and supply of different job profiles using different Natural Language Processing and Unsupervised Machine Learning techniques. The processing of the text data will be done by using an open source Language Model (LLM) in order to generate their corresponding document embeddings.