Models and Metrics for Mining Meaningful Metadata

Tyler J. Skluzacek; Matthew Chen; Erica Hsu; Kyle Chard; Ian Foster

doi:10.1007/978-3-031-08751-6_30

Models and Metrics for Mining Meaningful Metadata

Tyler J. Skluzacek, Matthew Chen, Erica Hsu, Kyle Chard, Ian Foster

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

The increasing volume and variety of science data has led to the creation of metadata extraction systems that automatically derive and synthesize relevant information from files. A critical component of metadata extraction systems is a mechanism for mapping extractors—lightweight tools to mine information from a particular file types—to each file in a repository. However, existing methods do little to address the heterogeneity and scale of science data, thereby leaving valuable data unextracted or wasting significant compute resources applying incorrect extractors to data. We construct an extractor scheduler that leverages file type identification (FTI) methods. We show that by training lightweight multi-label, multi-class statistical models on byte samples from files, we can correctly map 35% more extractors to files than by using libmagic. Further, we introduce a metadata quality toolkit to automatically assess the utility of extracted metadata.

Original language	English
Title of host publication	Computational Science - ICCS 2022, 22nd International Conference, Proceedings
Editors	Derek Groen, Clélia de Mulatier, Valeria V. Krzhizhanovskaya, Peter M.A. Sloot, Maciej Paszynski, Jack J. Dongarra
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	417-430
Number of pages	14
ISBN (Print)	9783031087509
DOIs	https://doi.org/10.1007/978-3-031-08751-6_30
State	Published - 2022
Externally published	Yes
Event	22nd Annual International Conference on Computational Science, ICCS 2022 - London, United Kingdom Duration: Jun 21 2022 → Jun 23 2022

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	13350 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	22nd Annual International Conference on Computational Science, ICCS 2022
Country/Territory	United Kingdom
City	London
Period	06/21/22 → 06/23/22

Funding

Acknowledgements. We gratefully acknowledge Takuya Kurihana (University of Chicago) for sharing his machine learning expertise. This work is supported in part by the National Science Foundation under Grants No. 2004894 and 1757970, and used resources of the Argonne Leadership Computing Facility. We gratefully acknowledge Takuya Kurihana (University of Chicago) for sharing his machine learning expertise. This work is supported in part by the National Science Foundation under Grants No. 2004894 and 1757970, and used resources of the Argonne Leadership Computing Facility.

Keywords

Extraction
File type identification
Metadata quality

Access to Document

10.1007/978-3-031-08751-6_30

Cite this

Skluzacek, T. J., Chen, M., Hsu, E., Chard, K., & Foster, I. (2022). Models and Metrics for Mining Meaningful Metadata. In D. Groen, C. de Mulatier, V. V. Krzhizhanovskaya, P. M. A. Sloot, M. Paszynski, & J. J. Dongarra (Eds.), Computational Science - ICCS 2022, 22nd International Conference, Proceedings (pp. 417-430). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13350 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-08751-6_30

Skluzacek, Tyler J. ; Chen, Matthew ; Hsu, Erica et al. / Models and Metrics for Mining Meaningful Metadata. Computational Science - ICCS 2022, 22nd International Conference, Proceedings. editor / Derek Groen ; Clélia de Mulatier ; Valeria V. Krzhizhanovskaya ; Peter M.A. Sloot ; Maciej Paszynski ; Jack J. Dongarra. Springer Science and Business Media Deutschland GmbH, 2022. pp. 417-430 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{50332b1993ac416dbb9fa6bf6cf48a6a,

title = "Models and Metrics for Mining Meaningful Metadata",

abstract = "The increasing volume and variety of science data has led to the creation of metadata extraction systems that automatically derive and synthesize relevant information from files. A critical component of metadata extraction systems is a mechanism for mapping extractors—lightweight tools to mine information from a particular file types—to each file in a repository. However, existing methods do little to address the heterogeneity and scale of science data, thereby leaving valuable data unextracted or wasting significant compute resources applying incorrect extractors to data. We construct an extractor scheduler that leverages file type identification (FTI) methods. We show that by training lightweight multi-label, multi-class statistical models on byte samples from files, we can correctly map 35\% more extractors to files than by using libmagic. Further, we introduce a metadata quality toolkit to automatically assess the utility of extracted metadata.",

keywords = "Extraction, File type identification, Metadata quality",

author = "Skluzacek, \{Tyler J.\} and Matthew Chen and Erica Hsu and Kyle Chard and Ian Foster",

note = "Publisher Copyright: {\textcopyright} 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 22nd Annual International Conference on Computational Science, ICCS 2022 ; Conference date: 21-06-2022 Through 23-06-2022",

year = "2022",

doi = "10.1007/978-3-031-08751-6\_30",

language = "English",

isbn = "9783031087509",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "417--430",

editor = "Derek Groen and \{de Mulatier\}, Cl{\'e}lia and Krzhizhanovskaya, \{Valeria V.\} and Sloot, \{Peter M.A.\} and Maciej Paszynski and Dongarra, \{Jack J.\}",

booktitle = "Computational Science - ICCS 2022, 22nd International Conference, Proceedings",

}

Skluzacek, TJ, Chen, M, Hsu, E, Chard, K & Foster, I 2022, Models and Metrics for Mining Meaningful Metadata. in D Groen, C de Mulatier, VV Krzhizhanovskaya, PMA Sloot, M Paszynski & JJ Dongarra (eds), Computational Science - ICCS 2022, 22nd International Conference, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 13350 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 417-430, 22nd Annual International Conference on Computational Science, ICCS 2022, London, United Kingdom, 06/21/22. https://doi.org/10.1007/978-3-031-08751-6_30

Models and Metrics for Mining Meaningful Metadata. / Skluzacek, Tyler J.; Chen, Matthew; Hsu, Erica et al.
Computational Science - ICCS 2022, 22nd International Conference, Proceedings. ed. / Derek Groen; Clélia de Mulatier; Valeria V. Krzhizhanovskaya; Peter M.A. Sloot; Maciej Paszynski; Jack J. Dongarra. Springer Science and Business Media Deutschland GmbH, 2022. p. 417-430 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 13350 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Models and Metrics for Mining Meaningful Metadata

AU - Skluzacek, Tyler J.

AU - Chen, Matthew

AU - Hsu, Erica

AU - Chard, Kyle

AU - Foster, Ian

PY - 2022

Y1 - 2022

N2 - The increasing volume and variety of science data has led to the creation of metadata extraction systems that automatically derive and synthesize relevant information from files. A critical component of metadata extraction systems is a mechanism for mapping extractors—lightweight tools to mine information from a particular file types—to each file in a repository. However, existing methods do little to address the heterogeneity and scale of science data, thereby leaving valuable data unextracted or wasting significant compute resources applying incorrect extractors to data. We construct an extractor scheduler that leverages file type identification (FTI) methods. We show that by training lightweight multi-label, multi-class statistical models on byte samples from files, we can correctly map 35% more extractors to files than by using libmagic. Further, we introduce a metadata quality toolkit to automatically assess the utility of extracted metadata.

AB - The increasing volume and variety of science data has led to the creation of metadata extraction systems that automatically derive and synthesize relevant information from files. A critical component of metadata extraction systems is a mechanism for mapping extractors—lightweight tools to mine information from a particular file types—to each file in a repository. However, existing methods do little to address the heterogeneity and scale of science data, thereby leaving valuable data unextracted or wasting significant compute resources applying incorrect extractors to data. We construct an extractor scheduler that leverages file type identification (FTI) methods. We show that by training lightweight multi-label, multi-class statistical models on byte samples from files, we can correctly map 35% more extractors to files than by using libmagic. Further, we introduce a metadata quality toolkit to automatically assess the utility of extracted metadata.

KW - Extraction

KW - File type identification

KW - Metadata quality

UR - https://www.scopus.com/pages/publications/85134336768

U2 - 10.1007/978-3-031-08751-6_30

DO - 10.1007/978-3-031-08751-6_30

M3 - Conference contribution

AN - SCOPUS:85134336768

SN - 9783031087509

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 417

EP - 430

BT - Computational Science - ICCS 2022, 22nd International Conference, Proceedings

A2 - Groen, Derek

A2 - de Mulatier, Clélia

A2 - Krzhizhanovskaya, Valeria V.

A2 - Sloot, Peter M.A.

A2 - Paszynski, Maciej

A2 - Dongarra, Jack J.

PB - Springer Science and Business Media Deutschland GmbH

T2 - 22nd Annual International Conference on Computational Science, ICCS 2022

Y2 - 21 June 2022 through 23 June 2022

ER -

Skluzacek TJ, Chen M, Hsu E, Chard K, Foster I. Models and Metrics for Mining Meaningful Metadata. In Groen D, de Mulatier C, Krzhizhanovskaya VV, Sloot PMA, Paszynski M, Dongarra JJ, editors, Computational Science - ICCS 2022, 22nd International Conference, Proceedings. Springer Science and Business Media Deutschland GmbH. 2022. p. 417-430. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-031-08751-6_30