Exploiting user models to avoid misconstruals
Andrea S. Hemerly, Marco A. Casanova, Antonio Luz Furtado · 1994
Abstract When interacting with a database, a user is typically tempted to infer further information from that explicitly obtained from previous queries. However, his inferences are not necessarily valid, because his model of the world is often incomplete or even faulty. This chapter addresses the problem of avoiding users’ faulty inferences or misconstruals. It describes a cooperative interface to deductive databases that alters the processing of users’ requests to include additional information that will block faulty inferences. The interface is governed by predefined user models, created during the design of the database. It also maintains a log of previous interactions with each user that records all information he has already acquired in his current session. The inferences of a given user are identified with those possible from the user model and the information in ,the log. The main result of the chapter says that, given any user model and any deductive database, all logs produced by the interface indeed contain sufficient information to block all misconstruals.