Bag-based data models for incomplete information and their closure properties

摘要

In most practical database applications, incompleteness and duplication of facts should be carefully handled. We propose bag-based data models for incomplete information, called V-bags, CV-bags, GV-bags, and CGV-bags. In V-bags, incompleteness is represented by variables like C-tables by Imielinski and Lipski. GV-bags are a supermodel of V-bags, where global conditions that restrict assignments over variables are attached. CV-bags and CGV-bags are submodels of V-bags and GV-bags, respectively, where the usage of variables for representing the number of duplication of tuples is somewhat restricted. We also investigate the closure properties of forward and inverse algebraic operations (selection, projection, product, union, difference, and unique) on each of the data models under both CWA and OWA. Among these data models, CGV-bags have the most closed operations.