LIMBO: Scalable Clustering of Categorical Data Export

@incollection{citeulike:4016480, abstract = {Clustering is a problem of great practical importance in numerous applications. The problem of clustering becomes more challenging when the data is categorical, that is, when there is no inherent distance measure between data values. We introduce LIMBO, a scalable hierarchical categorical clustering algorithm that builds on the Information Bottleneck (IB) framework for quantifying the relevant information preserved when clustering. As a hierarchical algorithm, LIMBO has the advantage that it can produce clusterings of different sizes in a single execution. We use the IB framework to define a distance measure for categorical tuples and we also present a novel distance measure for categorical attribute values. We show how the LIMBO algorithm can be used to cluster both tuples and values. LIMBO handles large data sets by producing a memory bounded summary model for the data. We present an experimental evaluation of LIMBO, and we study how clustering quality compares to other categorical clustering algorithms. LIMBO supports a trade-off between efficiency (in terms of space and time) and quality. We quantify this trade-off and demonstrate that LIMBO allows for substantial improvements in efficiency with negligible decrease in quality.}, author = {Andritsos, Periklis and Tsaparas, Panayiotis and Miller, Ren\'{e}e J. and Sevcik, Kenneth C.}, citeulike-article-id = {4016480}, citeulike-linkout-0 = {http://www.springerlink.com/content/bv76grdqnb2jlkg9}, comment = {* Input: 1 rel table * Output: Column matchings * Technique: \# Uses information theoretic measures to create a distance between tuples and between categorical values (categorical = no inherent distance measure exists) \# A hierarchical clustering algorithm (Agglomerative Information Bottleneck) uses then the measure to cluster tuples and/or values. \# The whole LIMBO clustering algo can be bounded in space and/or in time. Making it thus suitable for clustering streaming data. }, journal = {Advances in Database Technology - EDBT 2004}, keywords = {clustering, discovery, entityguides, schema}, pages = {531--532}, posted-at = {2009-02-06 15:09:48}, priority = {0}, title = {LIMBO: Scalable Clustering of Categorical Data}, url = {http://www.springerlink.com/content/bv76grdqnb2jlkg9}, year = {2004} }

TY - CHAP ID - citeulike:4016480 L3 - citeulike-article-id:4016480 N2 - Clustering is a problem of great practical importance in numerous applications. The problem of clustering becomes more challenging when the data is categorical, that is, when there is no inherent distance measure between data values. We introduce LIMBO, a scalable hierarchical categorical clustering algorithm that builds on the Information Bottleneck (IB) framework for quantifying the relevant information preserved when clustering. As a hierarchical algorithm, LIMBO has the advantage that it can produce clusterings of different sizes in a single execution. We use the IB framework to define a distance measure for categorical tuples and we also present a novel distance measure for categorical attribute values. We show how the LIMBO algorithm can be used to cluster both tuples and values. LIMBO handles large data sets by producing a memory bounded summary model for the data. We present an experimental evaluation of LIMBO, and we study how clustering quality compares to other categorical clustering algorithms. LIMBO supports a trade-off between efficiency (in terms of space and time) and quality. We quantify this trade-off and demonstrate that LIMBO allows for substantial improvements in efficiency with negligible decrease in quality. JF - Advances in Database Technology - EDBT 2004 EP - 532 TI - LIMBO: Scalable Clustering of Categorical Data SP - 531 KW - clustering KW - discovery KW - entityguides KW - schema N1 - * Input: 1 rel table * Output: Column matchings * Technique: # Uses information theoretic measures to create a distance between tuples and between categorical values (categorical = no inherent distance measure exists) # A hierarchical clustering algorithm (Agglomerative Information Bottleneck) uses then the measure to cluster tuples and/or values. # The whole LIMBO clustering algo can be bounded in space and/or in time. Making it thus suitable for clustering streaming data. AU - Andritsos, Periklis AU - Tsaparas, Panayiotis AU - Miller, Renée AU - Sevcik, Kenneth PY - 2004/// UR - http://www.springerlink.com/content/bv76grdqnb2jlkg9 ER -