3.4.9 Speech acts in TRIPS

Speech acts are used in TRIPS by the Interpretation Manager in interpreting the intentions of the user.

Speech acts are used in TRIPS by the Interpretation Manager in interpreting the intentions of the user. Intentions are then used, in turn, by the Task Manager in order to evaluate what the user's plan with regard to the task is. If there is more than one possible speech act assigned to an utterance, the right one with the deriving intention that better fits the task will be finally assigned to it. The user's plan of dealing with the task is then evaluated. Based on that plan evaluation the system can then take care of the necessary arrangements for the plans to be realised or suggest alternatives. This is the basis of the mixed initiative modeled by TRIPS.

The Interpretation Manager also uses speech acts for the manipulation of discourse obligations. Obligations represent the standard behaviour imposed by the domain. Based on these obligations, the system can disambiguate any ambiguous input. The Generation Manager, as well, decides on the content of the next system utterance by help of obligations. In general the building block of its reasoning are speech acts.

Adjacency pairs and insertions are also handled by use of speech acts so that there is no need to call the problem solving modules when dealing with these phenomena. For example, the system can automatically issue acknowledgements, that is, declarations of having understood what the user said, following a statement by the user.

On the whole the manipulation of discourse based on speech acts makes the separation of task and discourse planning possible. This separation results in a robust system and allows further improvements in each sector while the other one can be left intact.