Implementation of a pipeline in order to extract information from a transaction short description text: a knowledge discovery on card transaction data

This work aims to discover knowledge from a card transaction dataset. The starting data model is characterized by both structured data and unstructured data, thus we implement a pipeline for processing the transaction short description text in order to extract information by using a Named Entities Recognition (NER) model. Afterwards, we enrich the starting data model by extracted metadata and, using a dataset got from a real context, try to identify clusters among users.