There are several platforms and engines for building general-purpose chatbots. The application context must be considered when deciding which technologies to use. This work analyzed candidate scenarios for the development of educational chatbots through a comparative analysis using the scientific Benchmark methodology. Forty-three platforms were analyzed based on the details of the problem class, which pointed to the Rasa NLU framework as the most suitable. Therefore, this work offers the possibility of more assertive decision-making, exposing the desired characteristics for a chatbot in the context of online learning in Virtual Learning Environments.