Application of Artificial Intelligence Methods in a Word-Prediction Aid

Word-prediction appears to be a good aid to enhance message-composition rate for people with physical disabilities. Usually word-prediction is based on statistical information (mainly on word frequencies). Hit rate can be enhanced by trying to imitate the behaviour of a human interlocutor (who uses syntactic and semantic information). In this paper some new approaches based on Artificial Intelligence methods are presented. Advantages of syntactic and semantic analysis in relation to bare statistical methods are studied. Furthermore, the integration with human-computer interfaces for disabled users is also described.