Utilização de bases de conhecimento e word embeddings para enriquecimento dos dados de treinamento de chatbots conversacionais

Abstract

Dialog systems or chatbots are computer programs that use data to understand and answer questions asked by users. However, the chatbot does not always have this set of data ready to be used in its initial phases. In this way, the lack of initial data becomes a recurring issue in the creation of chatbots. This work aims to increase the chatbot input data by incorporating similar words and/or synonyms through the use of Word Embeddings and WordNet lexical knowledge base. This data augmentation is achieved through the development of an Application Programming Interface. The steps followed in the methodology begin with the definition of technologies, followed by the modeling and implementation of the data generator model, after which the implementation of the Natural Language Understanding is carried out. The fourth step is the case study and finally, the evaluation of the results of the experiments performed.