MyNewsWave: User-centered Web search and news delivery.
Clint Heyer, Jamie M. Madden, Kelly Hollingsworth, Peter Heydon, Keiran Bartlett, Joachim Diederich · 2002
MyNewsWave uses machine learning (including support vector machines) for a user-centred approach to full-text information retrieval as well as news delivery. The system uses knowledge sources such as WordNet to refine keyword queries and learns user-preferences with regard to web search. MyNewsWave includes an audio mining system for topic detection in conjunction with background search to facilitate the retrieval of relevant multimedia information. A special feature of MyNewsWave is the assessment of incoming information with regard to the “mood ” or personal relevance to a user. DigiMood is a component of MyNewsWave that classifies web pages into mood categories. Business news, for instance, can be classified by DigiMood to access market sentiment. Marconi analyses incoming news streams and uses machine learning to adjust parameters of a text-to-speech system. The objective is to learn the appropriate voice for news items as part of a speech user interface.