M-PIRO is concerned with the development of language engineering technology for Personalised Information Objects, concentrating on multilingual information delivery in virtual and other museum settings, where M-PIRO will deliver highly personalised descriptions of museum exhibits. M-PIRO is concentrating on the following key technologies:
multilingual generation from a single source
adaptive user modelling for personalised information presentation
improved speech synthesis through the use of domain constraints
generation of spoken and written messages from the same source
authoring of the single source through the use of symbolic authoring techniques
It is easy to provide information about digital objects, for example by clicking on the object in a browser. But this technology does not provide information which is tailored to the user's interests or preferences, or which takes into account previous exchanges. For that to happen, the information needs to be stored in a different, non-textual fashion, and turned into a spoken or written message in the language of the user, when needed. This generation process can take into account context and personal preferences of the user, so that it looks to the user as if the object has been stored all along as a Personalised Information Object. The main goal of the M-PIRO project is to develop the language engineering aspects of Personalised Information Objects, concentrating on multilingual information delivery in virtual and other museum settings.
The work will concentrate on developing the following key technologies:
Multilingual text generation from a single source. We want to be able to generate messages in three languages (English, Greek and Italian) from the same source representation. Project work will involve definition of the multilingual natural language generation system and the development or extension of grammars and other generation resources for each of the three languages.
Adaptive user modelling for tailored information presentation. Using natural language generation and concept-to-speech systems allows us to tailor the information, not just to the context of use and the history of the exchange, but also to the user's personal preferences. Project work will concentrate on developing models where users are assigned to appropriate clusters through simple questionnaires before the visit starts or by automatic profiling during the visit.
Better speech synthesis through the use of domain constraints. The use of synthesised speech is still problematic, particularly for applications where users do not have time to get used to the synthetic voice. We believe it is possible to make progress in this area by developing speech synthesis technology, which takes into account the constraints of the domain in which the synthesised speech has to be delivered. Project work will involve the collection of spoken language suitable to the domain of the application in each of the three languages and the further development of unit selection techniques for use in this domain.
Month 6: Initial spec of architecture and major components
Month 12: Preliminary web-based prototype
Month 24: Extended web-based prototype
Month 34: Prototype for enhanced virtual reality tour.
Funding SchemeCSC - Cost-sharing contracts
15310 Aghia Paraskevi Attikis
WC2E 8HA London