Using background knowledge in the aggregation of imprecise evidence in databases

摘要

Background knowledge of data is often available, arising from a concept hierarchy, as integrity constraints, from database integration, or from knowledge possessed by domain experts. Frequently databases contain incomplete data which we may represent using Dempster–Shafer mass functions to represent evidence contained in subsets of the domain. These subsets may be represented in the database as partial values which are derived from background knowledge using logic programming to re-engineer the database. For such a data model we develop an aggregation operator which calculates Dempster–Shafer mass functions and thus facilities decision making and knowledge discovery in databases.