EASY ASSISTANT: UMA FERRAMENTA DE AUXÍLIO PARA O DESENVOLVIMENTO DE CHATBOTS DE DOMÍNIO ESPECÍFICO

Intelligent systems are mainly linked to the act of bringing more and more support to people. Currently, there is a growing use of elements related to the machine learning capacity, that is, Artificial Intelligence (AI). In AI, chatbots stand out, personalized and virtual agents with a certain emotional and cognitive level, usually employed in collaborative, social or learning systems, in order to offer some service with persistent and strongly relational interaction to users. Developing such an agent is not trivial. A chatbot that provides consistent and immersive communication requires several techniques, sometimes even complex,related to natural language processing, understanding and generation, which involve classification of entities, objectives, among others; also self-healing to avoid conversation inconsistencies and loss of context. Therefore, this work presents a web tool that allows the user to develop, train and customize their chatbot virtual agent. This agent uses two training and response models based on recovery and generative models, built on neurais networks that manipulate a database defined by the user. The tool was tested and evaluated in order to qualify its accuracy, which proved to reach its goals, as well as allow the creation of a personalized chatbot.