| This thesis addresses the problem of using domain generalization graphs to generalize temporal data extracted from relational databases. Our approach is based on domain generalization graphs that define a partial order and encode domain knowledge relevant to generalization. We propose a grammar for domain generalization graphs and functioning specifications of calendar (date and time) and duration domain generalization graphs. Our proposed methods for adapting and pruning domain generalization graphs are explored along with methods for ranking and filtering the resulting generalized relations. We demonstrate the feasibility of our approach, that a single domain generalization graph can be used on multiple sets of data, and that the process provides summaries.; An interactive system for visualizing the generalization space allows the user to guide the data exploration process. The user is presented with a view of the Generalization Space that has some of the uninteresting generalizations automatically removed. In this way, the user is presented with generalized relations that embody real, previously unknown, useful, and understandable structure within the data. |
