Skip to Main Content
Presents an inter-class pattern discovery method for real-world databases. While the data in a conventional database has a tuple structure, the data in pattern discovery has set-values or sequences. The structural differences between them may cause useless resulting patterns, and may result in an inefficient pattern discovery method. To resolve those issues, we propose an inter-class pattern discovery methodology. The first step is to convert a conventional database into a set of objects. During the conversion process, a tuple in the original database is converted into a conceptual object and, as another result, object generalization hierarchies are generated. From the object generalization hierarchies, interesting patterns of the conceptual objects can be extracted by applying multi-level pattern discovery algorithms. The resulting patterns are inter-class patterns of the original conventional database. We also show a pattern discovery query for our methodology and its application to intelligent query processing.