Re-classification or building new ways to interpret text are part of a process called ‘custom modelling’ which involves a group ‘teaching’ the system about a new entity and relationship between entities