Soluções de Aprendizagem Automática e Inteligência Artificial para Interação Otimizada com Clientes e Utilizadores

Abstract

The described project addresses the execution of a data migration process from a relational database with historical documents to a graph database, creating the correct representation of the documents and associated entities. To model the data and create all the classes and relations in the graphs, the CIDOC-CRM ontology is used. The process leverages Machine Learning (ML) and Artificial Intelligence (AI) models to extract multiple entities, including people, places and organizations. Two ML tasks are performed: text information extraction to obtain the entities and create their representations in the graph database, and disambiguation of the extracted information to analyse the context in which the entity is involved, thus avoiding the creation of duplicate information or information loss. The goal is to develop a model capable of extracting the necessary information from the data in the relational database and migrating it to the graph database without losing any information or context. Additionally, an advanced search engine has been implemented to all the entities, with respective forms to create and edit all the data.