PPP-Codes for Large-Scale Similarity Searching

D 2016

PPP-Codes for Large-Scale Similarity Searching

NOVÁK, David and Pavel ZEZULA

Basic information

Original name

PPP-Codes for Large-Scale Similarity Searching

Authors

NOVÁK, David (203 Czech Republic, guarantor, belonging to the institution) and Pavel ZEZULA (203 Czech Republic, belonging to the institution)

Edition

Berlin Heidelberg, Transactions on Large-Scale Data- and Knowledge-Centered Systems XXIV, p. 61-87, 27 pp. 2016

Publisher

Springer

Other information

Language

English

Type of outcome

Stať ve sborníku

Field of Study

10201 Computer sciences, information science, bioinformatics

Country of publisher

Germany

Confidentiality degree

není předmětem státního či obchodního tajemství

Publication form

printed version "print"

References:

publisher link

Impact factor

Impact factor: 0.402 in 2005

RIV identification code

RIV/00216224:14330/16:00087959

Organization unit

Faculty of Informatics

ISBN

978-3-662-49213-0

ISSN

DOI

http://dx.doi.org/10.1007/978-3-662-49214-7_2

UT WoS

000452460000002

Keywords in English

similarity search; pivot permutations; approximate search; kNN search

Abstract

V originále

Many current applications need to organize data with respect to mutual similarity between data objects. A typical general strategy to retrieve objects similar to a given sample is to access and then refine a candidate set of objects. We propose an indexing and search technique that can significantly reduce the candidate set size by combination of several space partitionings. Specifically, we propose a mapping of objects from a generic metric space onto main memory codes using several pivot spaces; our search algorithm first ranks objects within each pivot space and then aggregates these rankings producing a candidate set reduced by two orders of magnitude while keeping the same answer quality. Our approach is designed to well exploit contemporary HW: (1) larger main memories allow us to use rich and fast index, (2) multi-core CPUs well suit our parallel search algorithm, and (3) SSD disks without mechanical seeks enable efficient selective retrieval of candidate objects. The gain of the significant candidate set reduction is paid by the overhead of the candidate ranking algorithm and thus our approach is more advantageous for datasets with expensive candidate set refinement, i.e. large data objects or expensive similarity function. On real-life datasets, the search time speedup achieved by our approach is by factor of two to five.

Links

GBP103/12/G084, research and development project

Name: Centrum pro multi-modální interpretaci dat velkého rozsahu

Investor: Czech Science Foundation

Citovat

NOVÁK, David and Pavel ZEZULA. PPP-Codes for Large-Scale Similarity Searching. In Abdelkader Hameurlain, Josef Küng, Roland Wagner, Hendrik Decker, Lenka Lhotska, Sebastian Link. Transactions on Large-Scale Data- and Knowledge-Centered Systems XXIV. Berlin Heidelberg: Springer, 2016, p. 61-87. ISBN 978-3-662-49213-0. Available from: https://dx.doi.org/10.1007/978-3-662-49214-7_2.

@inproceedings{1346582,
   author = {Novák, David and Zezula, Pavel},
   address = {Berlin Heidelberg},
   booktitle = {Transactions on Large-Scale Data- and Knowledge-Centered Systems XXIV},
   doi = {http://dx.doi.org/10.1007/978-3-662-49214-7_2},
   editor = {Abdelkader Hameurlain, Josef Küng, Roland Wagner, Hendrik Decker, Lenka Lhotska, Sebastian Link},
   keywords = {similarity search; pivot permutations; approximate search; kNN search},
   howpublished = {tištěná verze "print"},
   language = {eng},
   location = {Berlin Heidelberg},
   isbn = {978-3-662-49213-0},
   pages = {61-87},
   publisher = {Springer},
   title = {PPP-Codes for Large-Scale Similarity Searching},
   url = {http://link.springer.com/chapter/10.1007/978-3-662-49214-7_2},
   year = {2016}
}

TY  - JOUR
ID  - 1346582
AU  - Novák, David - Zezula, Pavel
PY  - 2016
TI  - PPP-Codes for Large-Scale Similarity Searching
PB  - Springer
CY  - Berlin Heidelberg
SN  - 9783662492130
KW  - similarity search
KW  - pivot permutations
KW  - approximate search
KW  - kNN search
UR  - http://link.springer.com/chapter/10.1007/978-3-662-49214-7_2
L2  - http://link.springer.com/chapter/10.1007/978-3-662-49214-7_2
N2  - Many current applications need to organize data with respect to mutual similarity between data objects. A typical general strategy to retrieve objects similar to a given sample is to access and then refine a candidate set of objects. We propose an indexing and search technique that can significantly reduce the candidate set size by combination of several space partitionings. Specifically, we propose a mapping of objects from a generic metric space onto main memory codes using several pivot spaces; our search algorithm first ranks objects within each pivot space and then aggregates these rankings producing a candidate set reduced by two orders of magnitude while keeping the same answer quality. Our approach is designed to well exploit contemporary HW: (1) larger main memories allow us to use rich and fast index, (2) multi-core CPUs well suit our parallel search algorithm, and (3) SSD disks without mechanical seeks enable efficient selective retrieval of candidate objects. The gain of the significant candidate set reduction is paid by the overhead of the candidate ranking algorithm and thus our approach is more advantageous for datasets with expensive candidate set refinement, i.e. large data objects or expensive similarity function. On real-life datasets, the search time speedup achieved by our approach is by factor of two to five.
ER  -

NOVÁK, David and Pavel ZEZULA. PPP-Codes for Large-Scale Similarity Searching. In Abdelkader Hameurlain, Josef Küng, Roland Wagner, Hendrik Decker, Lenka Lhotska, Sebastian Link. \textit{Transactions on Large-Scale Data- and Knowledge-Centered Systems XXIV}. Berlin Heidelberg: Springer, 2016, p.~61-87. ISBN~978-3-662-49213-0. Available from: https://dx.doi.org/10.1007/978-3-662-49214-7\_{}2.

Detailed Information on Publication Record