Email has been recognized as one of the chief sources of the current epidemic of information

overload. The work described in this paper is part of an ongoing project to re-focus the attention of

information workers on “what needs to be done” rather than “what someone wants us to read”. The

approach is to base a software agent that pre-processes incoming email on the user’s “personal

ontology”. The ontology used in this work contained two facets, namely Work Topics and Task Types,

and also included lexical “clues” whereby, if such a word or phrase is found in the text of a message,

the system can infer with a certain probability that a particular ontology concept (class or instance) is

relevant. The paper describes the prototype tools developed and some initial trials. Further trials and

improvements are planned.