HP Labs Technical Reports

Designing Contextual Information Filters

Paepcke, Andreas; Fafchamps, Danielle; Jeffries, Robin

HPL-91-145

Keyword(s):

Abstract: We propose that filter designers must consider not only filtering technology and user models in their designs but must also attempt to include the information about users' professional domain. This makes filters more likely to produce needed information at the right time. But it requires that filter designers get to know user work practices which are relevant to filtering. This understanding must often be gained by cooperative efforts with domain analysts. In order to support this cooperation, we suggest a simple framework for mentally organizing the actions of a computerized filter. We subdivide its operation into six intermediate steps and explain which domain information is needed for each step. We then outline an ethnographic approach for obtaining this information.

Back to Index