SUBJECT: NL&IP Seminar Tuesday 28th January Natural Language and Information Processing Group Seminar &WEBSITE 12pm , Tuesday 28th January , &NAME &ORG , &NAME &NAME Building , &NAME &NAME Avenue &WEBSITE &NAME : Document filtering : tasks , models and implementation Speaker : &NAME &NAME - &NAME Research &NAME : Much information retrieval research revolves around a canonical model of a search task : a user issuing a query to a ( temporarily static ) document collection , and being presented with a ranked list of documents ( with perhaps a feedback loop ) . This model informs many aspects of the work - - the kinds of theoretical approach taken , many implementation issues including the design of efficient , fast-response systems , the measurement of effectiveness and the qualitative testing of systems . However , we may take a number of different views of the information task itself and the user context in which that task occurs . &NUM alternative is represented by the filtering task , as defined for the &NAME exercise . Here filtering is seen as a content-based , adaptive task applied to an incoming stream of documents , involving user needs represented by profiles . In this talk , I will explore the nature of this task , and some of the ways in which it must be treated differently from the canonical model . I will address both conceptual and implementation issues , and discuss some experimental results from &NAME . ( All welcome - For instructions on how to get to the &ORG , please see : &WEBSITE