Hybrid record linkage model for integrating marine data

Devi Fitrianah; Ito Wasito

doi:10.1016/j.proeng.2012.10.100

Hybrid record linkage model for integrating marine data

Devi Fitrianah, Ito Wasito

Department of Computer Science

Research output: Contribution to journal › Conference article › peer-review

1 Citation (Scopus)

Abstract

The data integration from two different marine data sources is conducted with the hybrid record linkage model. The method is to find the same key attribute from each source and to decide whether the status is exact, mathed or unmatched. The Hybrid Record Linkage model is the combination between supervised learning and unsupervised learning. The study used set of 325 weekly composite Sea Surface Temperature and Chlorophyll a images derived from MODIS AQUA sensor and covering 4 years from 2007-2010 converted to ASCII form and Tuna Fishing Data from 2007-2010. The result from the study shows that the combination hybrid model GMM clustering and Naive Bayes Classifier identified more matched status and generate more accurate matched records with the average accuracy 0,950677.

Original language	English
Pages (from-to)	926-932
Number of pages	7
Journal	Procedia Engineering
Volume	50
DOIs	https://doi.org/10.1016/j.proeng.2012.10.100
Publication status	Published - 2012
Event	2012 International Conference on Advances Science and Contemporary Engineering, ICASCE 2012 - Jakarta, Indonesia Duration: 24 Oct 2012 → 25 Oct 2012

Keywords

Hybrid record linkage
Supervised learning
Unsupervised learning

Access to Document

10.1016/j.proeng.2012.10.100

Cite this

@article{aa75bca9ca0548b1ac0520b8118b9003,

title = "Hybrid record linkage model for integrating marine data",

abstract = "The data integration from two different marine data sources is conducted with the hybrid record linkage model. The method is to find the same key attribute from each source and to decide whether the status is exact, mathed or unmatched. The Hybrid Record Linkage model is the combination between supervised learning and unsupervised learning. The study used set of 325 weekly composite Sea Surface Temperature and Chlorophyll a images derived from MODIS AQUA sensor and covering 4 years from 2007-2010 converted to ASCII form and Tuna Fishing Data from 2007-2010. The result from the study shows that the combination hybrid model GMM clustering and Naive Bayes Classifier identified more matched status and generate more accurate matched records with the average accuracy 0,950677.",

keywords = "Hybrid record linkage, Supervised learning, Unsupervised learning",

author = "Devi Fitrianah and Ito Wasito",

year = "2012",

doi = "10.1016/j.proeng.2012.10.100",

language = "English",

volume = "50",

pages = "926--932",

journal = "Procedia Engineering",

issn = "1877-7058",

publisher = "Elsevier B.V.",

note = "2012 International Conference on Advances Science and Contemporary Engineering, ICASCE 2012 ; Conference date: 24-10-2012 Through 25-10-2012",

}

TY - JOUR

T1 - Hybrid record linkage model for integrating marine data

AU - Fitrianah, Devi

AU - Wasito, Ito

PY - 2012

Y1 - 2012

N2 - The data integration from two different marine data sources is conducted with the hybrid record linkage model. The method is to find the same key attribute from each source and to decide whether the status is exact, mathed or unmatched. The Hybrid Record Linkage model is the combination between supervised learning and unsupervised learning. The study used set of 325 weekly composite Sea Surface Temperature and Chlorophyll a images derived from MODIS AQUA sensor and covering 4 years from 2007-2010 converted to ASCII form and Tuna Fishing Data from 2007-2010. The result from the study shows that the combination hybrid model GMM clustering and Naive Bayes Classifier identified more matched status and generate more accurate matched records with the average accuracy 0,950677.

AB - The data integration from two different marine data sources is conducted with the hybrid record linkage model. The method is to find the same key attribute from each source and to decide whether the status is exact, mathed or unmatched. The Hybrid Record Linkage model is the combination between supervised learning and unsupervised learning. The study used set of 325 weekly composite Sea Surface Temperature and Chlorophyll a images derived from MODIS AQUA sensor and covering 4 years from 2007-2010 converted to ASCII form and Tuna Fishing Data from 2007-2010. The result from the study shows that the combination hybrid model GMM clustering and Naive Bayes Classifier identified more matched status and generate more accurate matched records with the average accuracy 0,950677.

KW - Hybrid record linkage

KW - Supervised learning

KW - Unsupervised learning

UR - http://www.scopus.com/inward/record.url?scp=84891815753&partnerID=8YFLogxK

U2 - 10.1016/j.proeng.2012.10.100

DO - 10.1016/j.proeng.2012.10.100

M3 - Conference article

AN - SCOPUS:84891815753

SN - 1877-7058

VL - 50

SP - 926

EP - 932

JO - Procedia Engineering

JF - Procedia Engineering

T2 - 2012 International Conference on Advances Science and Contemporary Engineering, ICASCE 2012

Y2 - 24 October 2012 through 25 October 2012

ER -

Hybrid record linkage model for integrating marine data

Abstract

Keywords

Access to Document

Other files and links

Fingerprint

Cite this