| Summary: | E-goi is an organization which provides automated multichannel marketing possibilities. Given its system’s complexity, it requires a not so smooth learning curve, which means that sometimes costumers incur upon some difficulties which directs them towards appropriate Costumer Support resources. With an increase in the number of users, these Costumer Support requests are somewhat frequent and demand an increase in availability in Costumer Support channels which become inundated with simple, easily-resolvable requests. The organization idealized the possibility of automating significant portion of costumer generated tickets with the possibility of scaling to deal with other types of operations. This thesis aims to present a long-term solution to that request with the development of a chatbot system, fully integrated with the existing enterprise modules and data sources. In order to accomplish this, prototypes using several Chatbot management and Natural Language Processing frameworks were developed. Afterwards, their advantages and disadvantages were pondered, followed by the implementation of its accompanying system and testing of developed software and Natural Language Processing results. Although the developed overarching system achieved its designed functionalities, the master’s thesis could not offer a viable solution for the problem at hand given that the available data could not provide an intent mining model usable in a real-world context.
|
|---|