Design de Usabilidade em Interfaces Conversacionais Híbridas

Abstract

This dissertation portrays essentially the usability design in Conversational Interfaces that allows the user to interact via natural language. Even though these interfaces evolve steadily and present a greater tendency in relation to accession by technology companies, there are still several problems and limitations when it comes to usability. Thus, the challenge of this dissertation involves the application of e-commerce Shelf AI, developed by Xarevision, which allows the user to make purchases, look for the product via text, voice or barcode. Although it is possible to search by voice, this does not include the ability for the user to interact through a natural dialogue. Having said that, the goal is to turn the Shelf AI into a hybrid conversational interface, where the user has the possibility to interact in hybrid form, i.e. by touch and voice/text. In the case of conversational interfaces, usability depends on mostly dialog flow, which is how the user interacts with the system through natural language. The design process focuses mainly on the dialog flow, where it is necessary to understand the user, a clear understanding of human factors as well as other factors that may interfere negatively with the usability of the system. Taking into account that the Shelf AI is being converted into a hybrid interface, new features were defined which led to the amendment or new modifications in relation to the graphic component as well as conversational.