Obtenção de dados estruturados para aquisição automática do domínio de planejamento gerenciador do agente de diálogo plantão coronavírus

Abstract

Automated Planning is a subfield of Artificial Intelligence (AI) that focuses on the deliberative process of selecting actions for an intelligent agent to achieve its goals. A planner is an algorithm capable of solving general problems, taking as input a high-level description of the agent and the environment (planning domain) and providing as output a sequence of actions (plan) that leads the agent from an initial state to a state that satisfies its goals. Acquiring a planning domain is a complex task, as it requires knowledge of action languages and the application environment, necessitating collaboration between planning specialists and domain experts. This process can be performed manually or automatically, using algorithms that learn the domain description from data, known as plan traces. This work aims to build a database of plan traces (structured data) from the policy generated by the PRP Planner, avoiding the need to extract data from natural language information. The policy used in this work was extracted through dialogues from the chatbot named Plantão Coronavírus, developed to assist the Government of the State of Ceará in communication between citizens and health professionals during the COVID-19 pandemic. By analyzing the policy generated by the PRP Planner, it was possible to more accurately obtain the start and end states for each action and to construct a solid database of plan traces, contributing to advances in the automatic acquisition of planning domains.