This research project aims at developing a portuguese dialogue system to interact with an online calendar application, the Google Calendar. The dialogue system uses the L2F NLP Chain in order to be able to automatically recognize and normalize temporal expressions. This thesis traces the entire development process, including: (i) the study of other dialogue systems that deal with the calendar domain; (ii) the study of dialogue systems developed based on the DIGA Framework; (iii) the design of a solution; (iv) the implementation and evaluation of the devised solution. In this work the DIGA Framework is used on a client perspective, in order to identify possible design and development problems and suggest improvements to the development of the framework.
This dialogue system supports four types of interaction with the calendar: scheduling, listing, deleting and searching of events in an online calendar associated with the user. Additionally, the support of a wide range of temporal expressions diminishes the constrains over the dialogue between users and the dialogue system. This work went through two evaluation phases which showed mixed results towards the user satisfaction. This work resulted in a proof of concept dialogue system which successfully recognizes temporal expressions in portuguese and is able to interact effectively with Google Calendar.