Architecture for Automatic Generation of User Interaction Guides with Intelligent Assistant

In recent years, Intelligent Assistant services have come out to respond to user requests or to perform certain functions. These services get the request input from users with natural ways such as voice or text. However, the biggest hurdle of this approach is that it is very difficult for users to know the scope of the service (feature or knowledge) and therefore it is difficult to deliver the right request. If the Intelligent Assistant suggests the proper features based on user's context or preference with users based on the natural language or voices, it could make users be familiar with Intelligent Assistant therefore it could make Intelligent Assistants be very widely used. In this paper, we proposed the architecture to make the user's interaction guide for the Intelligent Assistant. We used the plan recognition of the task network to make the candidates of the actions/answers based on the user's context. After making the candidates we used the filters to filtering the candidates and scored the candidates. Finally, we used the NLG/TTS to make the natural sentences to give the guide to the users. As a result, even without the user's request, we could give the guide utterance to the user.