Incorporating Uncertainty into Data Integration
Database systems typically model only certain data. That is, if a tuple is in the database, then it is true in the real world. There are many contexts in which our knowledge of the world is uncertain, and we would like to model that uncertainty in the database and be able to query it. For example, consider situations like the following:
• We are collecting information about theft sightings, and a witness does not remember whether the thief had blond or brown hair.
• We are recording temperature measurements from sensors in the field, and the accuracy of the sensors has a certain error bound or measurements may even be completely wrong if a sensor’s battery is low.
• We are integrating data across ...