Deep Learning for social services
With the collaboration of Universidad Pompeu Fabra (Barcelona), Eticas carried out an audit of the natural language processing (NLP) system from the Social Services area of the Barcelona City Council.
The NLP system was developed to categorize the transcribed verbal content of interviews performed with citizens by social workers.
Purpose
The purpose of the data treatment was to classify the nature of different citizens’ demands into defined groups in order to streamline resource assignment for the population’s needs, reducing the human effort required for allocation.
Our contribution
This collaborative study between Eticas and the university focused on two main goals:
Accuracy Evaluation: Assessing how accurately the deep learning model classified demands, problems, and resources based on interview data collected by social workers.
Fairness and Impact Analysis; Determining whether the system could have a negatively differentiated impact on socially disadvantaged groups due to incorrect or biased associations in certain types of demands, problems, or resources.