Exploring and Comparing Table Fragments With Fragment Summaries

摘要

In this article, we introduce a new pattern-based summarization framework for representing and reasoning about fragmented data sets. A fragment summary is a concise, complete and precise representation of a data fragment and its information contents relatively to the whole data set. We formally define the notion of fragment summary and the use of Structured Query Language (SQL) queries over fragment summaries for analyzing data fragments. We introduce an algorithm for computing summaries and present an experimental evaluation using two real-life data sets.