The Object Summary (OS) is a recently proposed
tree structure, which summarizes all data held in a relational database about a
data subject. An OS can potentially be very large in size and therefore
unfriendly for users who wish to view synoptic information about the data
subject. In this paper, we investigate the effective and efficient retrieval of
concise and informative OS snippets (denoted as size-l OSs). We propose and
investigate the effectiveness of two types of size- l OSs, namely size- l OS
(t)s and size-l OS (a)s that consist of l tuple nodes and l attribute nodes
respectively. For computing size-l OSs, we propose an optimal dynamic
programming algorithm, two greedy algorithms and preprocessing heuristics. By
collecting feedback from real users (e.g., from DBLP authors), we assess the
relative usability of the two different types of snippets, the choice of the
size- l parameter, as well as the effectiveness of the snippets with respect to
the user expectations. In addition, via thorough evaluation on real databases,
we test the speed and effectiveness of our techniques.
No comments:
Post a Comment