The document discusses domain classification and word sense disambiguation systems. It describes a domain classifier that uses support vector machines to assign domain labels to texts from 37 predefined domains. It also describes three word sense disambiguation systems: timbl-DSC which uses k-nearest neighbor classification, svm-DSC which uses support vector machines, and ukb-DSC which is an unsupervised knowledge-based system. The systems are evaluated using fold cross-validation, random evaluation on texts from SONAR, and evaluation on independent texts, with the combination of systems achieving the best performance.