Dialog modeling of domain specific cognitive assistants

Resumen

Intelligent virtual assistants of software are becoming more common. Frequently, the communication interface is dialogue which is expressed in natural language. When an assistant consists of questions that are associated with a single answer, the abstraction of the dialogue is straightforward, however, when there are conversations with multiple bifurcations in the flow of communication, the representation and documentation is complex. In this sense, there is a lack of formalities that allow a dialogue to be accurately represented and based on them, to proceed with its programming in an industrial framework. In this work we propose the use of finite state machines to model complex dialogues. Thus, we introduce a formal model for the representation of dialogue and its mapping to a real assistant, using an industrial framework. We believe that our approach will make it easier to design and to program assistants with complex interaction, enabling better adoption of them.