useR! 2024

International standard classifications such as ISCO (for Occupation), ISCED (for Education) and COICOP (for Consumption) serve as pivotal statistical frameworks for the organization and classification of information. In official statistical practices, adherence to these codes is essential for thorough analysis and comparison of findings. Survey respondents typically provide information in an unstructured free textual format, requiring subsequent assignment to standardized code. This process is often done manually, resulting in time-consuming laborious tasks. In our talk, we propose an approach that automates the classification of textual data into various standardized codes using simple mathematical techniques combined with neural network-based language models, utilizing the R libraries TensorFlow and Keras. Additionally, we illustrate the development of application programming interfaces (APIs) using plumber, and the deployment of our models through posit connect, establishing accessibility to a broad user base.