Technical Reports

A List by Author: Stanislav Bartoò

e-mail:

xbarton(a)fi.muni.cz

Adaptive Approximate Similarity Searching through Metric Social Networks

by Jan Sedmidubský, Stanislav Bartoò, Vlastislav Dohnal, Pavel Zezula, A full version of the paper presented at ICDE 2008. November 2007, 22 pages.

FIMU-RS-2007-06. Available as Postscript, PDF.

Abstract:

Exploiting the concepts of social networking represents a novel approach to the approximate similarity query processing. We present an unstructured and dynamic P2P environment in which a metric social network is built. Social communities of peers giving similar results to specific queries are established and such ties are exploited for answering future queries. Based on the universal law of generalization, a new query forwarding algorithm is introduced and evaluated. The same principle is used to manage query histories of individual peers with the possibility to tune the tradeoff between the extent of the history and the level of the query-answer approximation. All proposed algorithms are tested on real data and medium-sized P2P networks consisting of tens of computers.

rhoIndex, Designing and Evaluating an Indexing Structure for Graph Structured Data

by Stanislav Bartoò, Pavel Zezula, A full version of the paper presented at IEEE MCD 2006. September 2006, 22 pages.

FIMU-RS-2006-07. Available as Postscript, PDF.

Abstract:

An own design of an indexing structure for general graph structured data called rhoIndex that allows an effective processing of special path queries is presented. These special queries represent for example a search for all paths lying between two arbitrary vertices limited to a certain path length. The rhoIndex is a multilevel balanced tree structure where each node is created with a certain graph transformation and described by modified adjacency matrix. Hence, rhoIndex indexes all the paths to a predefined length linclusive. The search algorithm is then able to find all the paths shorter than or having the length land some of the paths longer then the predefined llying between any two vertices in the indexed graph. The designed search algorithm exploits a special graph structure, a transcription graph, to compute the result using the rhoIndex . We also present an experimental evaluation of the process of creating the rhoIndex on graphs with different sizes and also a complexity evaluation of the search algorithm that uses the rhoIndex.