TreeCluster: Clustering results of keyword search over databases

Publication Type:
Journal Article
Citation:
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2006, 4016 LNCS pp. 385 - 396
Issue Date:
2006-01-01
Metrics:
Full metadata record
Files in This Item:
Filename Description Size
Thumbnail2012007543OK.pdf512.13 kB
Adobe PDF
A critical challenge in keyword search over relational databases (KSORD) is to improve its result presentation to facilitate users' quick browsing through search results. An effective method is to organize the results into clusters. However, traditional clustering method is not applicable to KSORD search results. In this paper, we propose a novel clustering method named TreeCluster. In the first step, we use labels to represent schema information of each result tree and reformulate the clustering problem as a problem of judging whether labeled trees are isomorphic. In the second step, we rank user keywords according to their frequencies in databases, and further partition the large clusters based on keyword nodes. Furthermore, we give each cluster a readable description, and present the description and each result graphically to help users understand the results more easily. Experimental results verify our method's effectiveness and efficiency. © Springer-Verlag Berlin Heidelberg 2006.
Please use this identifier to cite or link to this item: