Speaking the Users' Languages

interests, or preferences, or that takes account of previous exchanges between the user and the system. For that to happen, the information must be stored in a different, nontextual fashion and turned into a spoken or written message in the user’s language when needed. This generation process can take into account the context and the user’s linguistic and personal preferences so that it looks to the user as if the object has been stored all along as a personalized information object. Usability expert Jakob Nielsen has long argued that computer systems should speak the user’s language. By using personalized information objects, a system can indeed speak the user’s language. Moreover, a multilingual system can speak all the users’ languages. Such objects will find application in many different arenas. For instance, GPS-equipped alwayson mobile devices will offer location-based services, providing users with information about nearby objects and organizations. These include both commercial and cultural products and providers: sales offers, restaurants, historical artifacts, and tourist attractions. But the attractiveness and utility of location-based services could be greatly enhanced if they trade specifically in personalized information objects. Users expect customized information and they will expect it to be delivered using whatever presentation method makes most sense: graphics, text, speech, or all three. We don’t need to wait for next-generation mobile devices to explore the possibilities of personalized information objects. The system we have been working on, within the M-PIRO project, has concentrated on virtual and other museum settings. A specific advantage is that museum communication professionals are well-versed in the need to tailor an organization’s preferred message to their visitors’expertise and information needs. The project has focused on developing language-engineering technology for personalized information objects, specifically on multilingual information delivery. Progress on the project has required working on multilingual natural-language generation from a single source; improving speech synthesis for languages other than English; generating spoken and written messages from the same source; authoring the single source through symbolic authoring techniques; and working with adaptive user modeling for personalized information presentation. The result is a stateof-the-art system that speaks the users’ languages and offers highly personalized information objects.