Incremental Generalization for Mining in a Data Warehousing Environment

摘要: On a data warehouse, either manual analyses supported by appropriate visualization tools or (semi-) automatic mining may be performed, e.g. clustering, classification and summarization. Attribute-oriented generalization is common method for the task of Typically, in warehouse update operations are collected applied to periodically. Then, all derived information has updated as well. Due very large size base relations, it highly desirable perform these updates incrementally. In this paper, we present algorithms incremental attribute-oriented with conflicting goals good efficiency minimal overly generalization. The insertions deletions based on materialization relation at an intermediate level, i.e. anchor relation. Our experiments demonstrate that can performed efficiently low degree Furthermore, optimal cardinality sets determined experimentally yielding best efficiency.